diff --git a/src/application/use-cases/manage-bucket.ts b/src/application/use-cases/manage-bucket.ts
new file mode 100644
index 0000000..a51e0b9
--- /dev/null
+++ b/src/application/use-cases/manage-bucket.ts
@@ -0,0 +1,168 @@
+import type { Bucket } from '../../domain/entities/bucket';
+import type { IBucketRepository } from '../../domain/ports/bucket-repository';
+import type { IFileRepository } from '../../domain/ports/file-repository';
+
+/**
+ * S3 bucket name validation regex.
+ *
+ * Bucket names must be 3-63 characters, start/end with a lowercase letter or
+ * digit, and contain only lowercase letters, digits, dots, and hyphens.
+ */
+const BUCKET_NAME_REGEX = /^[a-z0-9][a-z0-9.-]{1,61}[a-z0-9]$/;
+
+/**
+ * Error type for bucket-level application errors that carry an S3-compatible
+ * error code and an HTTP status suggestion.
+ */
+export class BucketError extends Error {
+  /** S3-compatible error code (e.g. "NoSuchBucket", "BucketAlreadyExists"). */
+  readonly code: string;
+  /** Suggested HTTP status code. */
+  readonly status: number;
+
+  /**
+   * @param code - The S3 error code.
+   * @param message - Human-readable error description.
+   * @param status - Suggested HTTP status.
+   */
+  constructor(code: string, message: string, status: number) {
+    super(message);
+    this.name = 'BucketError';
+    this.code = code;
+    this.status = status;
+  }
+}
+
+/**
+ * Describes a bucket entry returned by the list-buckets use case,
+ * enriched with the current count of non-deleted objects.
+ */
+export interface BucketWithCount {
+  /** The bucket domain entity. */
+  bucket: Bucket;
+  /** Number of non-deleted objects in the bucket. */
+  objectCount: number;
+}
+
+/** Dependencies required by the manage-bucket use case factories. */
+export interface ManageBucketDeps {
+  /** Bucket repository for CRUD operations. */
+  bucketRepo: IBucketRepository;
+  /** File repository for counting and checking objects within buckets. */
+  fileRepo: IFileRepository;
+}
+
+/**
+ * Creates a use case that lists all buckets together with their object counts.
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function that returns a list of buckets with counts.
+ */
+export function createListBucketsUseCase(deps: ManageBucketDeps) {
+  return async (): Promise<BucketWithCount[]> => {
+    const buckets = await deps.bucketRepo.list();
+    const results = await Promise.all(
+      buckets.map(async (bucket) => ({
+        bucket,
+        objectCount: await deps.fileRepo.countByBucket(bucket.id),
+      })),
+    );
+    return results;
+  };
+}
+
+/**
+ * Creates a use case that retrieves a single bucket by its name.
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting a bucket name and returning the
+ *          bucket, or `null` when not found.
+ */
+export function createGetBucketUseCase(deps: ManageBucketDeps) {
+  return async (name: string): Promise<Bucket | null> => {
+    return deps.bucketRepo.findByName(name);
+  };
+}
+
+/**
+ * Creates a use case that creates a new bucket.
+ *
+ * Validates the bucket name format and checks for duplicates before
+ * persisting.
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting a bucket name and returning the
+ *          newly created bucket.
+ * @throws {BucketError} When the name is invalid or the bucket already exists.
+ */
+export function createCreateBucketUseCase(deps: ManageBucketDeps) {
+  return async (name: string): Promise<Bucket> => {
+    if (!BUCKET_NAME_REGEX.test(name)) {
+      throw new BucketError(
+        'InvalidBucketName',
+        'The specified bucket is not valid.',
+        400,
+      );
+    }
+
+    const existing = await deps.bucketRepo.findByName(name);
+    if (existing) {
+      throw new BucketError(
+        'BucketAlreadyExists',
+        'The requested bucket name is not available.',
+        409,
+      );
+    }
+
+    return deps.bucketRepo.create(name);
+  };
+}
+
+/**
+ * Creates a use case that deletes a bucket.
+ *
+ * Ensures the bucket exists and is empty (no non-deleted objects) before
+ * proceeding with deletion.
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting a bucket name. Returns `true` when
+ *          the bucket was deleted, throws when the bucket is missing or
+ *          not empty.
+ * @throws {BucketError} When the bucket does not exist or is not empty.
+ */
+export function createDeleteBucketUseCase(deps: ManageBucketDeps) {
+  return async (name: string): Promise<boolean> => {
+    const bucket = await deps.bucketRepo.findByName(name);
+    if (!bucket) {
+      throw new BucketError(
+        'NoSuchBucket',
+        'The specified bucket does not exist.',
+        404,
+      );
+    }
+
+    const objectCount = await deps.fileRepo.countByBucket(bucket.id);
+    if (objectCount > 0) {
+      throw new BucketError(
+        'BucketNotEmpty',
+        'The bucket you tried to delete is not empty.',
+        409,
+      );
+    }
+
+    return deps.bucketRepo.delete(name);
+  };
+}
+
+/**
+ * Creates a use case that checks whether a bucket exists.
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting a bucket name and returning `true`
+ *          when the bucket exists.
+ */
+export function createBucketExistsUseCase(deps: ManageBucketDeps) {
+  return async (name: string): Promise<boolean> => {
+    return deps.bucketRepo.exists(name);
+  };
+}
\ No newline at end of file
diff --git a/src/application/use-cases/multipart-upload.ts b/src/application/use-cases/multipart-upload.ts
new file mode 100644
index 0000000..4bf8148
--- /dev/null
+++ b/src/application/use-cases/multipart-upload.ts
@@ -0,0 +1,378 @@
+import { nanoid } from 'nanoid';
+import type { MultipartUpload } from '../../domain/entities/multipart';
+import type { IBucketRepository } from '../../domain/ports/bucket-repository';
+import type { IFileRepository } from '../../domain/ports/file-repository';
+import type { IMultipartRepository } from '../../domain/ports/multipart-repository';
+import type { ITelegramService } from '../../domain/ports/telegram-service';
+import { computeHash } from '../../shared/utils/file';
+
+// ─── Types ──────────────────────────────────────────────────────────
+
+/**
+ * A single part reference as submitted in a complete-multipart-upload request.
+ */
+export interface CompletePartInput {
+  /** 1-based part number. */
+  partNumber: number;
+  /** ETag returned when the part was uploaded. */
+  etag: string;
+}
+
+/**
+ * Result of initiating a multipart upload.
+ */
+export interface InitiateMultipartResult {
+  /** The generated upload identifier (nanoid). */
+  uploadId: string;
+  /** The bucket name. */
+  bucket: string;
+  /** The S3 object key. */
+  key: string;
+}
+
+/**
+ * Result of uploading a single part.
+ */
+export interface UploadPartResult {
+  /** ETag of the uploaded part (SHA-256 hex digest). */
+  etag: string;
+}
+
+/**
+ * Result of completing a multipart upload.
+ */
+export interface CompleteMultipartResult {
+  /** Public-facing unique identifier of the created file record. */
+  publicId: string;
+  /** The S3 location URL of the completed object. */
+  location: string;
+  /** Combined ETag (all part etags joined by hyphens). */
+  etag: string;
+  /** Total object size in bytes. */
+  sizeBytes: number;
+}
+
+/**
+ * Summary of a single part within a multipart upload used in listing results.
+ */
+export interface PartSummary {
+  /** 1-based part number. */
+  partNumber: number;
+  /** ETag of the part content. */
+  etag: string;
+  /** Part size in bytes. */
+  sizeBytes: number;
+  /** ISO-8601 timestamp when the part was stored. */
+  createdAt: Date;
+}
+
+/**
+ * Result of listing multipart uploads within a bucket.
+ */
+export interface ListMultipartUploadsResult {
+  /** Array of in-progress upload summaries. */
+  uploads: MultipartUpload[];
+  /** Whether more results are available. */
+  isTruncated: boolean;
+  /** Marker for the next page, or null when not truncated. */
+  nextKeyMarker: string | null;
+}
+
+// ─── Config ─────────────────────────────────────────────────────────
+
+/** Subset of application configuration consumed by the multipart use cases. */
+export interface MultipartConfig {
+  /** Maximum chunk size in bytes for Telegram uploads (part size limit). */
+  telegramChunkSizeBytes: number;
+  /** Telegram chat ID where part data is stored. */
+  storageChatId: number;
+  /** Server base URL for constructing location URLs. */
+  baseUrl: string;
+}
+
+/** Dependencies required by the multipart upload use case factories. */
+export interface MultipartDeps {
+  /** Bucket repository for bucket lookups. */
+  bucketRepo: IBucketRepository;
+  /** File repository for creating the final file record on completion. */
+  fileRepo: IFileRepository;
+  /** Multipart repository for managing upload sessions and parts. */
+  multipartRepo: IMultipartRepository;
+  /** Telegram service for forwarding part data to storage. */
+  telegramService: ITelegramService;
+  /** Application configuration subset. */
+  config: MultipartConfig;
+}
+
+// ─── Use Case Factories ─────────────────────────────────────────────
+
+/**
+ * Creates a use case that initiates an S3 multipart upload.
+ *
+ * Validates the bucket exists and creates a new multipart upload session.
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting bucket name and object key, returning
+ *          the upload initiation result, or `null` when the bucket is not found.
+ */
+export function createInitiateMultipartUploadUseCase(deps: MultipartDeps) {
+  return async (
+    bucketName: string,
+    key: string,
+  ): Promise<InitiateMultipartResult | null> => {
+    const bucket = await deps.bucketRepo.findByName(bucketName);
+    if (!bucket) return null;
+
+    const uploadId = await deps.multipartRepo.create(bucket.id, key, 's3');
+
+    return { uploadId, bucket: bucketName, key };
+  };
+}
+
+/**
+ * Creates a use case that uploads a single part of a multipart upload.
+ *
+ * Validates the part number range (1-10000), checks the upload session exists
+ * and matches the expected key, checks part size against the configured limit,
+ * forwards the part data to Telegram storage, and persists the part record.
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting upload details and part data, returning
+ *          the part ETag, or `null` when the upload session is not found.
+ */
+export function createUploadPartUseCase(deps: MultipartDeps) {
+  return async (input: {
+    /** Bucket name for the multipart upload. */
+    bucketName: string;
+    /** S3 object key for the multipart upload. */
+    key: string;
+    /** Upload identifier returned by initiate. */
+    uploadId: string;
+    /** 1-based part number (1-10000). */
+    partNumber: number;
+    /** Raw part data. */
+    body: Buffer;
+  }): Promise<UploadPartResult | null> => {
+    if (input.partNumber < 1 || input.partNumber > 10000) {
+      throw new MultipartError(
+        'InvalidArgument',
+        'Part number must be an integer between 1 and 10000',
+        400,
+      );
+    }
+
+    const multipart = await deps.multipartRepo.findById(input.uploadId);
+    if (!multipart || multipart.s3Key !== input.key) {
+      return null;
+    }
+
+    if (input.body.byteLength > deps.config.telegramChunkSizeBytes) {
+      throw new MultipartError(
+        'EntityTooLarge',
+        `Your proposed upload part size (${input.body.byteLength} bytes) exceeds the maximum allowed part size (${deps.config.telegramChunkSizeBytes} bytes) for this storage backend. Use smaller part sizes.`,
+        400,
+      );
+    }
+
+    const forwardResult = await deps.telegramService.forwardToStorage(
+      input.body,
+      `mp-${input.uploadId}-part-${input.partNumber}`,
+      'document',
+    );
+
+    const etag = computeHash(input.body);
+
+    await deps.multipartRepo.insertPart({
+      uploadId: input.uploadId,
+      partNumber: input.partNumber,
+      telegramFileId: forwardResult.telegramFileId,
+      telegramFileUniqueId: forwardResult.telegramFileUniqueId,
+      storageMessageId: forwardResult.storageMessageId,
+      sizeBytes: input.body.byteLength,
+      etag,
+    });
+
+    return { etag: `"${etag}"` };
+  };
+}
+
+/**
+ * Error type for multipart-level application errors.
+ */
+export class MultipartError extends Error {
+  /** S3-compatible error code. */
+  readonly code: string;
+  /** Suggested HTTP status code. */
+  readonly status: number;
+
+  /**
+   * @param code - The S3 error code.
+   * @param message - Human-readable error description.
+   * @param status - Suggested HTTP status.
+   */
+  constructor(code: string, message: string, status: number) {
+    super(message);
+    this.name = 'MultipartError';
+    this.code = code;
+    this.status = status;
+  }
+}
+
+/**
+ * Creates a use case that completes an S3 multipart upload.
+ *
+ * Validates the submitted part list (all parts must be present and in ascending
+ * order), creates the final file record referencing the first part's Telegram
+ * data, marks the upload session as completed, and returns the combined result.
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting upload details and submitted parts,
+ *          returning the completion result, or `null` when the upload session
+ *          is not found.
+ */
+export function createCompleteMultipartUploadUseCase(deps: MultipartDeps) {
+  return async (input: {
+    /** Bucket name for the multipart upload. */
+    bucketName: string;
+    /** S3 object key for the multipart upload. */
+    key: string;
+    /** Upload identifier. */
+    uploadId: string;
+    /** Parts submitted by the client (in ascending part number order). */
+    parts: CompletePartInput[];
+  }): Promise<CompleteMultipartResult | null> => {
+    const multipart = await deps.multipartRepo.findById(input.uploadId);
+    if (!multipart) return null;
+
+    const storedParts = await deps.multipartRepo.listParts(input.uploadId);
+
+    // Validate ascending part order
+    const partNumbers = input.parts.map((p) => p.partNumber);
+    if (partNumbers.length > 1 && partNumbers.some((n, i) => i > 0 && n <= partNumbers[i - 1])) {
+      throw new MultipartError(
+        'InvalidPartOrder',
+        'The list of parts was not in ascending order.',
+        400,
+      );
+    }
+
+    // Validate part count matches
+    if (input.parts.length !== storedParts.length) {
+      throw new MultipartError(
+        'InvalidPart',
+        'One or more specified parts could not be found.',
+        400,
+      );
+    }
+
+    const totalSize = storedParts.reduce((sum, p) => sum + p.sizeBytes, 0);
+    const firstPart = storedParts[0];
+    if (!firstPart) {
+      throw new MultipartError('InternalError', 'Multipart object has no parts.', 500);
+    }
+
+    const publicId = nanoid();
+
+    await deps.fileRepo.create({
+      publicId,
+      telegramFileId: firstPart.telegramFileId,
+      telegramFileUniqueId: firstPart.telegramFileUniqueId,
+      storageChatId: deps.config.storageChatId,
+      storageMessageId: firstPart.storageMessageId,
+      fileName: input.key.split('/').pop() || 'file',
+      mimeType: 'application/octet-stream',
+      sizeBytes: totalSize,
+      fileType: 'document',
+      uploaderId: 0,
+      fileHash: null,
+      archiveTelegramFileId: null,
+      archiveStorageMessageId: null,
+      archiveFileName: null,
+      archiveEntryName: null,
+      archiveMimeType: null,
+      archiveSizeBytes: null,
+      bucketId: multipart.bucketId,
+      s3Key: input.key,
+      storageBackend: 'telegram',
+      isDeleted: false,
+      multipartUploadId: input.uploadId,
+      partCount: null,
+    });
+
+    await deps.multipartRepo.complete(input.uploadId);
+
+    const location = `${deps.config.baseUrl}/${input.bucketName}/${input.key}`;
+    const combinedEtag = storedParts.map((p) => p.etag).join('-');
+
+    return {
+      publicId,
+      location,
+      etag: combinedEtag,
+      sizeBytes: totalSize,
+    };
+  };
+}
+
+/**
+ * Creates a use case that aborts an S3 multipart upload.
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting an upload identifier, returning `true`
+ *          when the upload was aborted, or `null` when the upload session
+ *          is not found.
+ */
+export function createAbortMultipartUploadUseCase(deps: MultipartDeps) {
+  return async (uploadId: string): Promise<boolean | null> => {
+    const multipart = await deps.multipartRepo.findById(uploadId);
+    if (!multipart) return null;
+
+    await deps.multipartRepo.abort(uploadId);
+    return true;
+  };
+}
+
+/**
+ * Creates a use case that lists in-progress multipart uploads within a bucket.
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting query parameters and returning the
+ *          listing result, or `null` when the bucket is not found.
+ */
+export function createListMultipartUploadsUseCase(deps: MultipartDeps) {
+  return async (input: {
+    /** Bucket name to list uploads from. */
+    bucketName: string;
+    /** Maximum number of uploads to return (clamped 1-1000). */
+    maxUploads: number;
+    /** Return only uploads whose S3 key is strictly greater than this, or null. */
+    keyMarker: string | null;
+  }): Promise<ListMultipartUploadsResult | null> => {
+    const bucket = await deps.bucketRepo.findByName(input.bucketName);
+    if (!bucket) return null;
+
+    return deps.multipartRepo.listByBucket(bucket.id, input.maxUploads, input.keyMarker);
+  };
+}
+
+/**
+ * Creates a use case that lists parts of a specific multipart upload.
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting an upload identifier and returning the
+ *          list of parts, or `null` when the upload session is not found.
+ */
+export function createListPartsUseCase(deps: MultipartDeps) {
+  return async (uploadId: string): Promise<PartSummary[] | null> => {
+    const multipart = await deps.multipartRepo.findById(uploadId);
+    if (!multipart) return null;
+
+    const parts = await deps.multipartRepo.listParts(uploadId);
+
+    return parts.map((p) => ({
+      partNumber: p.partNumber,
+      etag: p.etag,
+      sizeBytes: p.sizeBytes,
+      createdAt: p.createdAt,
+    }));
+  };
+}
\ No newline at end of file
diff --git a/src/application/use-cases/s3-object.ts b/src/application/use-cases/s3-object.ts
new file mode 100644
index 0000000..d125404
--- /dev/null
+++ b/src/application/use-cases/s3-object.ts
@@ -0,0 +1,774 @@
+import { randomUUID } from 'node:crypto';
+import { gzipSync } from 'node:zlib';
+import { nanoid } from 'nanoid';
+import type { File, NewFile } from '../../domain/entities/file';
+import type { NewFilePart } from '../../domain/entities/file-part';
+import type { MultipartPart } from '../../domain/entities/multipart';
+import type { IBucketRepository } from '../../domain/ports/bucket-repository';
+import type { IFilePartRepository } from '../../domain/ports/file-part-repository';
+import type { IFileRepository, S3FileRecord } from '../../domain/ports/file-repository';
+import type { IMultipartRepository } from '../../domain/ports/multipart-repository';
+import type { ITelegramService, TelegramFileInfo } from '../../domain/ports/telegram-service';
+import { ensureExtension, computeHash, formatCreatedAt } from '../../shared/utils/file';
+
+// ─── Types ──────────────────────────────────────────────────────────
+
+/**
+ * Compression algorithm used for chunked object storage.
+ */
+type CompressionAlgorithm = 'gzip' | null;
+
+/**
+ * A single part source for building a multi-part streaming response.
+ * Each part corresponds to a Telegram-stored file chunk.
+ */
+export interface ObjectPartSource {
+  /** Telegram file identifier for retrieving this part. */
+  telegramFileId: string;
+  /** Telegram CDN URL for downloading this part. */
+  telegramUrl: string;
+  /** Original size of this part in bytes. */
+  sizeBytes: number;
+  /** 1-based part number within the object. */
+  partNumber: number;
+  /** Stored (post-compression) size in bytes, when applicable. */
+  storedSizeBytes?: number;
+  /** Compression algorithm applied, or null if uncompressed. */
+  compressionAlgorithm?: CompressionAlgorithm;
+}
+
+/**
+ * A regular (direct) S3 object resolved to a Telegram CDN URL.
+ */
+export interface DirectObjectResult {
+  /** Discriminant. */
+  type: 'direct';
+  /** The resolved file entity. */
+  file: File;
+  /** Full Telegram CDN URL for downloading the object. */
+  telegramUrl: string;
+  /** Telegram file metadata. */
+  fileInfo: TelegramFileInfo;
+}
+
+/**
+ * A chunked S3 object stored across multiple Telegram file parts.
+ */
+export interface ChunkedObjectResult {
+  /** Discriminant. */
+  type: 'chunked';
+  /** The resolved file entity. */
+  file: File;
+}
+
+/**
+ * An S3 object assembled from a completed multipart upload.
+ */
+export interface MultipartObjectResult {
+  /** Discriminant. */
+  type: 'multipart';
+  /** The resolved file entity. */
+  file: File;
+  /** Resolved part sources with Telegram CDN URLs. */
+  parts: ObjectPartSource[];
+}
+
+/**
+ * Discriminated union of all possible S3 get-object outcomes.
+ */
+export type GetObjectResult = DirectObjectResult | ChunkedObjectResult | MultipartObjectResult;
+
+/**
+ * Result of an S3 put-object operation.
+ */
+export interface PutObjectResult {
+  /** SHA-256 hex digest of the object content. */
+  etag: string;
+}
+
+/**
+ * Result of an S3 copy-object operation.
+ */
+export interface CopyObjectResult {
+  /** SHA-256 hex digest of the source object content. */
+  etag: string;
+  /** ISO-8601 timestamp of the copy operation. */
+  lastModified: string;
+}
+
+/**
+ * A single S3 object as returned in listing results.
+ */
+export interface ListObjectEntry {
+  /** The object key (full path within the bucket). */
+  key: string;
+  /** Object size in bytes. */
+  sizeBytes: number;
+  /** SHA-256 hex digest or fallback identifier. */
+  etag: string;
+  /** ISO-8601 timestamp of last modification. */
+  lastModified: string;
+  /** MIME type of the stored object. */
+  mimeType: string;
+}
+
+/**
+ * Result of an S3 list-objects operation (both V1 and V2).
+ */
+export interface ListObjectsResult {
+  /** Array of object summaries. */
+  objects: ListObjectEntry[];
+  /** Common prefixes when a delimiter was used. */
+  prefixes: string[];
+  /** Whether more results are available. */
+  isTruncated: boolean;
+  /** The last key in the returned page, for use as the next marker. */
+  nextMarker: string | null;
+}
+
+/**
+ * Head-object metadata.
+ */
+export interface HeadObjectMetadata {
+  /** MIME type of the object. */
+  contentType: string;
+  /** Object size in bytes. */
+  contentLength: number;
+  /** Entity tag (SHA-256 hex digest or fallback). */
+  etag: string;
+  /** ISO-8601 timestamp of last modification. */
+  lastModified: string;
+}
+
+// ─── Config ─────────────────────────────────────────────────────────
+
+/** Subset of application configuration consumed by the s3-object use cases. */
+export interface S3ObjectConfig {
+  /** Maximum chunk size in bytes for Telegram chunked uploads. */
+  telegramChunkSizeBytes: number;
+  /** Whether gzip compression is enabled for chunked uploads. */
+  compressChunkedUploads: boolean;
+  /** Minimum chunk size in bytes below which compression is skipped. */
+  chunkCompressionMinSizeBytes: number;
+  /** Telegram chat ID where file parts are stored. */
+  storageChatId: number;
+  /** Server base URL for constructing download links. */
+  baseUrl: string;
+  /** Whether to proxy S3 GET requests through the server. */
+  proxyS3Get: boolean;
+}
+
+/** Dependencies required by the s3-object use case factories. */
+export interface S3ObjectDeps {
+  /** Bucket repository for bucket lookups. */
+  bucketRepo: IBucketRepository;
+  /** File repository for object CRUD operations. */
+  fileRepo: IFileRepository;
+  /** File-part repository for chunked upload part records. */
+  filePartRepo: IFilePartRepository;
+  /** Multipart repository for resolving multipart-upload objects. */
+  multipartRepo: IMultipartRepository;
+  /** Telegram service for uploading and resolving file metadata. */
+  telegramService: ITelegramService;
+  /** Application configuration subset. */
+  config: S3ObjectConfig;
+}
+
+// ─── Helpers ────────────────────────────────────────────────────────
+
+/**
+ * Validates the configured chunk size and returns it as a safe integer.
+ *
+ * @param chunkSizeBytes - The configured chunk size in bytes.
+ * @returns The same value if it is a positive safe integer.
+ */
+const asSafeChunkSize = (chunkSizeBytes: number): number => {
+  if (!Number.isSafeInteger(chunkSizeBytes) || chunkSizeBytes <= 0) {
+    throw new Error('Invalid Telegram chunk size');
+  }
+  return chunkSizeBytes;
+};
+
+/**
+ * Optionally gzip-compresses a chunk if compression is enabled and the chunk
+ * is large enough to benefit from it.
+ *
+ * @param chunk - The raw chunk buffer.
+ * @param compress - Whether compression is enabled.
+ * @param compressionMinSizeBytes - Minimum chunk size to attempt compression.
+ * @returns The (possibly compressed) buffer and the algorithm used.
+ */
+const maybeCompressChunk = (
+  chunk: Buffer,
+  compress: boolean,
+  compressionMinSizeBytes: number,
+): { bytes: Buffer; compressionAlgorithm: CompressionAlgorithm } => {
+  if (!compress || chunk.byteLength < compressionMinSizeBytes) {
+    return { bytes: chunk, compressionAlgorithm: null };
+  }
+
+  const gzipped = gzipSync(chunk);
+  if (gzipped.byteLength >= chunk.byteLength) {
+    return { bytes: chunk, compressionAlgorithm: null };
+  }
+
+  return { bytes: gzipped, compressionAlgorithm: 'gzip' };
+};
+
+/**
+ * Metadata for a single uploaded chunk/part during S3 put-object.
+ */
+interface UploadedPart {
+  /** 1-based part number. */
+  partNumber: number;
+  /** Telegram file identifier for this part. */
+  telegramFileId: string;
+  /** Telegram unique file identifier (stable across bot tokens). */
+  telegramFileUniqueId: string;
+  /** Message ID within the storage chat. */
+  storageMessageId: number;
+  /** Original size of the chunk in bytes before compression. */
+  sizeBytes: number;
+  /** Stored (post-compression) size in bytes. */
+  storedSizeBytes: number;
+  /** Compression algorithm applied, or null if uncompressed. */
+  compressionAlgorithm: CompressionAlgorithm;
+  /** SHA-256 hash of the original chunk content. */
+  etag: string;
+}
+
+/** Result of uploading an object in multiple Telegram chunks. */
+interface ChunkedUploadResult {
+  /** Metadata for each uploaded part. */
+  parts: UploadedPart[];
+  /** SHA-256 hex digest of the complete object content. */
+  fileHash: string;
+  /** Total object size in bytes (sum of all original chunks). */
+  totalSizeBytes: number;
+}
+
+/**
+ * Uploads a buffer to Telegram in chunks, returning metadata for all parts.
+ *
+ * @param buffer - The full object buffer.
+ * @param partFileNamePrefix - Prefix used for each chunk's file name in Telegram.
+ * @param chunkSizeBytes - Maximum size of each chunk in bytes.
+ * @param compress - Whether gzip compression is enabled.
+ * @param compressionMinSizeBytes - Minimum chunk size to attempt compression.
+ * @param telegramService - The Telegram service to forward each chunk.
+ * @returns The aggregated chunked upload result.
+ */
+const uploadInChunks = async (
+  buffer: Buffer,
+  partFileNamePrefix: string,
+  chunkSizeBytes: number,
+  compress: boolean,
+  compressionMinSizeBytes: number,
+  telegramService: ITelegramService,
+): Promise<ChunkedUploadResult> => {
+  const safeChunkSize = asSafeChunkSize(chunkSizeBytes);
+  const hasher = new Bun.CryptoHasher('sha256');
+  const parts: UploadedPart[] = [];
+  let totalSizeBytes = 0;
+  let partNumber = 0;
+  let offset = 0;
+
+  while (offset < buffer.byteLength) {
+    const chunk = buffer.subarray(offset, offset + safeChunkSize);
+    if (chunk.byteLength === 0) break;
+
+    partNumber += 1;
+    totalSizeBytes += chunk.byteLength;
+    hasher.update(chunk);
+
+    const { bytes, compressionAlgorithm } = maybeCompressChunk(
+      chunk,
+      compress,
+      compressionMinSizeBytes,
+    );
+
+    const forwardResult = await telegramService.forwardToStorage(
+      bytes,
+      `${partFileNamePrefix}.part-${partNumber}`,
+      'document',
+    );
+
+    parts.push({
+      partNumber,
+      telegramFileId: forwardResult.telegramFileId,
+      telegramFileUniqueId: forwardResult.telegramFileUniqueId,
+      storageMessageId: forwardResult.storageMessageId,
+      sizeBytes: chunk.byteLength,
+      storedSizeBytes: bytes.byteLength,
+      compressionAlgorithm,
+      etag: computeHash(chunk),
+    });
+
+    offset += safeChunkSize;
+  }
+
+  return {
+    parts,
+    fileHash: hasher.digest('hex'),
+    totalSizeBytes,
+  };
+};
+
+/**
+ * Resolves a list of multipart parts to their Telegram CDN URLs.
+ *
+ * @param parts - The stored multipart parts.
+ * @param telegramService - The Telegram service for resolving file metadata.
+ * @returns An array of resolved part sources.
+ */
+const resolveMultipartParts = async (
+  parts: MultipartPart[],
+  telegramService: ITelegramService,
+): Promise<ObjectPartSource[]> => {
+  const sources: ObjectPartSource[] = [];
+  for (const part of parts) {
+    const fileInfo = await telegramService.getFileInfo(part.telegramFileId);
+    sources.push({
+      telegramFileId: part.telegramFileId,
+      telegramUrl: `https://api.telegram.org/file/bot${fileInfo.bot_token}/${fileInfo.file_path}`,
+      sizeBytes: part.sizeBytes,
+      partNumber: part.partNumber,
+    });
+  }
+  return sources;
+};
+
+/**
+ * Formats a `createdAt` value into an HTTP Last-Modified header value.
+ *
+ * @param date - The date to format.
+ * @returns The UTC string representation.
+ */
+const formatLastModified = (date: Date | string | number): string => {
+  return date instanceof Date ? date.toUTCString() : new Date(date).toUTCString();
+};
+
+// ─── Use Case Factories ─────────────────────────────────────────────
+
+/**
+ * Creates a use case that resolves an S3 object for retrieval (GET).
+ *
+ * Looks up the bucket and file by key, then determines the storage type:
+ * - **direct**: regular Telegram-stored object — resolves the Telegram CDN URL.
+ * - **chunked**: object stored across multiple Telegram file parts.
+ * - **multipart**: object assembled from a completed multipart upload — resolves
+ *   the Telegram CDN URLs for each part.
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting bucket name and object key, returning
+ *          a discriminated union of possible results, or `null` when the
+ *          bucket or file is not found.
+ */
+export function createGetObjectUseCase(deps: S3ObjectDeps) {
+  return async (bucketName: string, key: string): Promise<GetObjectResult | null> => {
+    const bucket = await deps.bucketRepo.findByName(bucketName);
+    if (!bucket) return null;
+
+    const file = await deps.fileRepo.findByBucketAndKey(bucket.id, key);
+    if (!file) return null;
+
+    // Chunked storage — return the entity; the caller resolves parts via
+    // chunked-storage helpers.
+    if (file.storageBackend === 'chunked') {
+      return { type: 'chunked', file };
+    }
+
+    // Multipart upload object — resolve part Telegram URLs
+    if (file.multipartUploadId) {
+      const parts = await deps.multipartRepo.listParts(file.multipartUploadId);
+      const resolvedParts = await resolveMultipartParts(parts, deps.telegramService);
+      return { type: 'multipart', file, parts: resolvedParts };
+    }
+
+    // Regular direct object — resolve Telegram CDN URL
+    const fileInfo = await deps.telegramService.getFileInfo(file.telegramFileId);
+    const telegramUrl = `https://api.telegram.org/file/bot${fileInfo.bot_token}/${fileInfo.file_path}`;
+
+    return { type: 'direct', file, telegramUrl, fileInfo };
+  };
+}
+
+/**
+ * Creates a use case that retrieves S3 object metadata (HEAD).
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting bucket name and object key, returning
+ *          metadata or `null` when the bucket or file is not found.
+ */
+export function createHeadObjectUseCase(deps: S3ObjectDeps) {
+  return async (bucketName: string, key: string): Promise<HeadObjectMetadata | null> => {
+    const bucket = await deps.bucketRepo.findByName(bucketName);
+    if (!bucket) return null;
+
+    const file = await deps.fileRepo.findByBucketAndKey(bucket.id, key);
+    if (!file) return null;
+
+    return {
+      contentType: file.mimeType,
+      contentLength: file.sizeBytes,
+      etag: file.fileHash || nanoid(16),
+      lastModified: formatLastModified(file.createdAt),
+    };
+  };
+}
+
+/**
+ * Creates a use case that stores an S3 object (PUT).
+ *
+ * Handles both chunked (large) and single-message (small) upload paths,
+ * deduplicates by bucket+key (idempotent PUT), and persists the file
+ * record and (for chunked storage) part records.
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting bucket name, key, body buffer, and
+ *          content type, returning the etag of the stored object. Returns
+ *          `null` when the bucket is not found.
+ */
+export function createPutObjectUseCase(deps: S3ObjectDeps) {
+  return async (
+    bucketName: string,
+    key: string,
+    body: Buffer,
+    contentType: string,
+  ): Promise<PutObjectResult | null> => {
+    const bucket = await deps.bucketRepo.findByName(bucketName);
+    if (!bucket) return null;
+
+    const hash = computeHash(body);
+
+    // Idempotent PUT: if the object already exists, skip upload
+    const existing = await deps.fileRepo.findByBucketAndKey(bucket.id, key);
+    if (existing) {
+      return { etag: `"${hash}"` };
+    }
+
+    const fileName = key.split('/').pop() || 'file';
+    const signatureBuffer = body.subarray(0, 16);
+    const { fileName: finalFileName, mimeType } = ensureExtension(
+      fileName,
+      signatureBuffer,
+      contentType,
+    );
+
+    const partFileNamePrefix = `s3-${bucket.name}-${key.replace(/\//g, '_')}`;
+    const { telegramChunkSizeBytes, compressChunkedUploads, chunkCompressionMinSizeBytes, storageChatId } = deps.config;
+
+    if (body.byteLength > telegramChunkSizeBytes) {
+      // Chunked upload path
+      const chunkResult = await uploadInChunks(
+        body,
+        partFileNamePrefix,
+        telegramChunkSizeBytes,
+        compressChunkedUploads,
+        chunkCompressionMinSizeBytes,
+        deps.telegramService,
+      );
+
+      const firstPart = chunkResult.parts[0];
+      if (!firstPart) {
+        throw new Error('Chunked upload produced no parts');
+      }
+
+      const fileId = randomUUID();
+      const publicId = nanoid();
+
+      await deps.fileRepo.create({
+        publicId,
+        telegramFileId: firstPart.telegramFileId,
+        telegramFileUniqueId: firstPart.telegramFileUniqueId,
+        storageChatId,
+        storageMessageId: firstPart.storageMessageId,
+        fileName: finalFileName,
+        mimeType,
+        sizeBytes: chunkResult.totalSizeBytes,
+        fileType: 'document',
+        uploaderId: 0,
+        fileHash: chunkResult.fileHash,
+        archiveTelegramFileId: null,
+        archiveStorageMessageId: null,
+        archiveFileName: null,
+        archiveEntryName: null,
+        archiveMimeType: null,
+        archiveSizeBytes: null,
+        bucketId: bucket.id,
+        s3Key: key,
+        storageBackend: 'chunked',
+        isDeleted: false,
+        multipartUploadId: null,
+        partCount: chunkResult.parts.length,
+      });
+
+      const fileParts: NewFilePart[] = chunkResult.parts.map((part) => ({
+        fileId,
+        partNumber: part.partNumber,
+        telegramFileId: part.telegramFileId,
+        telegramFileUniqueId: part.telegramFileUniqueId,
+        storageChatId,
+        storageMessageId: part.storageMessageId,
+        sizeBytes: part.sizeBytes,
+        storedSizeBytes: part.storedSizeBytes,
+        compressionAlgorithm: part.compressionAlgorithm,
+        etag: part.etag,
+      }));
+
+      await deps.filePartRepo.insert(fileParts);
+
+      return { etag: `"${chunkResult.fileHash}"` };
+    }
+
+    // Single-message upload path
+    const forwardResult = await deps.telegramService.forwardToStorage(
+      body,
+      partFileNamePrefix,
+      'document',
+    );
+
+    const publicId = nanoid();
+
+    await deps.fileRepo.create({
+      publicId,
+      telegramFileId: forwardResult.telegramFileId,
+      telegramFileUniqueId: forwardResult.telegramFileUniqueId,
+      storageChatId,
+      storageMessageId: forwardResult.storageMessageId,
+      fileName: finalFileName,
+      mimeType,
+      sizeBytes: body.byteLength,
+      fileType: 'document',
+      uploaderId: 0,
+      fileHash: hash,
+      archiveTelegramFileId: null,
+      archiveStorageMessageId: null,
+      archiveFileName: null,
+      archiveEntryName: null,
+      archiveMimeType: null,
+      archiveSizeBytes: null,
+      bucketId: bucket.id,
+      s3Key: key,
+      storageBackend: 'telegram',
+      isDeleted: false,
+      multipartUploadId: null,
+      partCount: null,
+    });
+
+    return { etag: `"${hash}"` };
+  };
+}
+
+/**
+ * Creates a use case that copies an S3 object to a new key (PUT with
+ * x-amz-copy-source).
+ *
+ * Creates a new file record referencing the same Telegram-stored data.
+ * Chunked source objects are not supported for copy.
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting source + destination identifiers and
+ *          optional precondition headers, returning the copy result or
+ *          `null` when a required bucket or file is not found.
+ */
+export function createCopyObjectUseCase(deps: S3ObjectDeps) {
+  return async (input: {
+    /** Source bucket name. */
+    sourceBucket: string;
+    /** Source object key. */
+    sourceKey: string;
+    /** Destination bucket UUID (must already exist). */
+    destBucketId: string;
+    /** Destination object key. */
+    destKey: string;
+    /** Optional if-match precondition (raw etag value, without surrounding quotes). */
+    ifMatch?: string | null;
+    /** Optional if-none-match precondition (raw etag value, without surrounding quotes). */
+    ifNoneMatch?: string | null;
+  }): Promise<CopyObjectResult | null> => {
+    const sourceBucket = await deps.bucketRepo.findByName(input.sourceBucket);
+    if (!sourceBucket) return null;
+
+    const sourceFile = await deps.fileRepo.findByBucketAndKey(sourceBucket.id, input.sourceKey);
+    if (!sourceFile) return null;
+
+    if (sourceFile.storageBackend === 'chunked') {
+      throw new ObjectError('NotImplemented', 'Copying chunked objects is not yet implemented.', 501);
+    }
+
+    // Conditional copy: if-match / if-none-match checks
+    const sourceEtag = sourceFile.fileHash;
+    if (input.ifMatch && sourceEtag && input.ifMatch !== sourceEtag) {
+      throw new ObjectError('PreconditionFailed', 'The preconditions you specified did not hold.', 412);
+    }
+    if (input.ifNoneMatch && sourceEtag && input.ifNoneMatch === sourceEtag) {
+      throw new ObjectError('PreconditionFailed', 'The preconditions you specified did not hold.', 412);
+    }
+
+    const publicId = nanoid();
+
+    await deps.fileRepo.create({
+      publicId,
+      telegramFileId: sourceFile.telegramFileId,
+      telegramFileUniqueId: sourceFile.telegramFileUniqueId,
+      storageChatId: sourceFile.storageChatId,
+      storageMessageId: sourceFile.storageMessageId,
+      fileName: sourceFile.fileName,
+      mimeType: sourceFile.mimeType,
+      sizeBytes: sourceFile.sizeBytes,
+      fileType: sourceFile.fileType,
+      uploaderId: 0,
+      fileHash: sourceFile.fileHash,
+      archiveTelegramFileId: sourceFile.archiveTelegramFileId,
+      archiveStorageMessageId: sourceFile.archiveStorageMessageId,
+      archiveFileName: sourceFile.archiveFileName,
+      archiveEntryName: sourceFile.archiveEntryName,
+      archiveMimeType: sourceFile.archiveMimeType,
+      archiveSizeBytes: sourceFile.archiveSizeBytes,
+      bucketId: input.destBucketId,
+      s3Key: input.destKey,
+      storageBackend: 'telegram',
+      isDeleted: false,
+      multipartUploadId: null,
+      partCount: null,
+    });
+
+    return {
+      etag: sourceEtag || nanoid(16),
+      lastModified: new Date().toISOString(),
+    };
+  };
+}
+
+/**
+ * Error type for S3 object-level application errors.
+ */
+export class ObjectError extends Error {
+  /** S3-compatible error code. */
+  readonly code: string;
+  /** Suggested HTTP status code. */
+  readonly status: number;
+
+  /**
+   * @param code - The S3 error code.
+   * @param message - Human-readable error description.
+   * @param status - Suggested HTTP status.
+   */
+  constructor(code: string, message: string, status: number) {
+    super(message);
+    this.name = 'ObjectError';
+    this.code = code;
+    this.status = status;
+  }
+}
+
+/**
+ * Creates a use case that soft-deletes an S3 object (DELETE).
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting bucket name and object key, returning
+ *          `true` if a row was soft-deleted. Returns `null` when the bucket
+ *          is not found.
+ */
+export function createDeleteObjectUseCase(deps: S3ObjectDeps) {
+  return async (bucketName: string, key: string): Promise<boolean | null> => {
+    const bucket = await deps.bucketRepo.findByName(bucketName);
+    if (!bucket) return null;
+    return deps.fileRepo.softDelete(bucket.id, key);
+  };
+}
+
+/**
+ * Creates a use case that batch-deletes multiple S3 objects (POST with
+ * ?delete).
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting bucket name and an array of keys,
+ *          returning the array of keys that were actually deleted. Returns
+ *          `null` when the bucket is not found.
+ */
+export function createDeleteObjectsUseCase(deps: S3ObjectDeps) {
+  return async (bucketName: string, keys: string[]): Promise<string[] | null> => {
+    const bucket = await deps.bucketRepo.findByName(bucketName);
+    if (!bucket) return null;
+
+    const deletedKeys: string[] = [];
+    for (const key of keys) {
+      const ok = await deps.fileRepo.softDelete(bucket.id, key);
+      if (ok) deletedKeys.push(key);
+    }
+    return deletedKeys;
+  };
+}
+
+/**
+ * Creates a use case that lists objects within a bucket (ListObjectsV1/V2).
+ *
+ * Supports prefix filtering, delimiter-based pseudo-directory grouping, and
+ * pagination via marker/startAfter.
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting query parameters and returning the
+ *          listing result, or `null` when the bucket is not found.
+ */
+export function createListObjectsUseCase(deps: S3ObjectDeps) {
+  return async (input: {
+    /** Bucket name to list from. */
+    bucketName: string;
+    /** Key prefix to filter by (empty string for no filter). */
+    prefix: string;
+    /** Delimiter character (e.g. "/") or null for flat listing. */
+    delimiter: string | null;
+    /** Maximum number of object records to return (clamped to 1000). */
+    maxKeys: number;
+    /** Return only keys strictly greater than this value, or null. */
+    startAfter: string | null;
+  }): Promise<ListObjectsResult | null> => {
+    const bucket = await deps.bucketRepo.findByName(input.bucketName);
+    if (!bucket) return null;
+
+    const clampedMaxKeys = Math.min(input.maxKeys, 1000);
+
+    const { objects, prefixes } = await deps.fileRepo.listByPrefix(
+      bucket.id,
+      input.prefix,
+      input.delimiter,
+      clampedMaxKeys,
+      input.startAfter,
+    );
+
+    const isTruncated = objects.length > clampedMaxKeys;
+    const displayObjects = objects.slice(0, clampedMaxKeys);
+    const nextMarker = isTruncated
+      ? (displayObjects[displayObjects.length - 1]?.s3Key ?? null)
+      : null;
+
+    return {
+      objects: displayObjects.map((o: S3FileRecord) => ({
+        key: o.s3Key,
+        sizeBytes: o.sizeBytes,
+        etag: o.fileHash || nanoid(16),
+        lastModified: formatCreatedAt(o.createdAt),
+        mimeType: o.mimeType,
+      })),
+      prefixes,
+      isTruncated,
+      nextMarker,
+    };
+  };
+}
+
+/**
+ * Creates a use case that checks whether an object exists and is accessible
+ * within a bucket.
+ *
+ * @param deps - The injected dependencies.
+ * @returns An async function accepting a bucket ID and object key,
+ *          returning the file entity or null.
+ */
+export function createFindObjectUseCase(deps: Pick<S3ObjectDeps, 'fileRepo'>) {
+  return async (bucketId: string, key: string): Promise<File | null> => {
+    return deps.fileRepo.findByBucketAndKey(bucketId, key);
+  };
+}
\ No newline at end of file