diff --git a/cdmtaskservice/job_state.py b/cdmtaskservice/job_state.py
index 711f118..2bee0c4 100644
--- a/cdmtaskservice/job_state.py
+++ b/cdmtaskservice/job_state.py
@@ -310,16 +310,20 @@ async def stream_job_logs(
         job_id: str,
         container_num: int,
         user: CTSUser,
-        stderr: bool = False
+        stderr: bool = False,
+        seek: int | None = None,
+        length: int | None = None,
     ) -> tuple[AsyncIterator[bytes], str]:
         """
         Stream the container logs from a job.
-        
+
         job_id - the job's ID.
         container_num - the container number for which to retrieve logs.
         user - the user requesting the logs.
         stderr - return the stderr logs instead of the stdout logs.
-        
+        seek - the byte offset in the file from which to start reading.
+        length - the number of bytes to read from the file.
+
         Returns a tuple of a generator that will stream the logfile and the name of the file.
         """
         job = await self.get_job(job_id, user)
@@ -332,7 +336,25 @@ async def stream_job_logs(
             )
         filename = s3errpath if stderr else s3outpath
         s3path = S3Paths([str(Path(job.logpath) / filename)])
-        return self._s3.stream_object(s3path), filename
+
+        if seek is not None:
+            if seek < 0:
+                raise IllegalParameterError(
+                    f"Seek parameter must be >= 0, got {seek}"
+                )
+            # Only fetch file metadata if seek is non-zero
+            if seek > 0:
+                meta = await self._s3.get_object_meta(s3path)
+                if seek >= meta[0].size:
+                    raise IllegalParameterError(
+                        f"Seek parameter {seek} is >= file size {meta[0].size}"
+                    )
+        if length is not None and length < 1:
+            raise IllegalParameterError(
+                f"Length parameter must be >= 1, got {length}"
+            )
+
+        return self._s3.stream_object(s3path, seek=seek, length=length), filename
     
     async def resend_job_notification(self, job_id: str, user: CTSUser, state: models.JobState):
         """
diff --git a/cdmtaskservice/routes.py b/cdmtaskservice/routes.py
index f49cd4d..d4970b6 100644
--- a/cdmtaskservice/routes.py
+++ b/cdmtaskservice/routes.py
@@ -344,6 +344,14 @@ async def get_job_exit_codes(
     description="The container / subjob number.",
     ge=0,
 )]
+_ANN_LOG_SEEK = Annotated[int, Query(
+    description="The byte offset in the file from which to start reading.",
+    ge=0,
+)]
+_ANN_LOG_LENGTH = Annotated[int, Query(
+    description="The number of bytes to read from the file.",
+    ge=1,
+)]
 
 
 @ROUTER_JOBS.get(
@@ -360,8 +368,10 @@ async def get_job_stdout(
     job_id: _ANN_JOB_ID,
     container_num: _ANN_CONTAINER_NUMBER,
     user: CTSUser=Depends(_AUTH),
+    seek: _ANN_LOG_SEEK = None,
+    length: _ANN_LOG_LENGTH = None,
 ) -> StreamingResponse:
-    return await get_logs(r, job_id, container_num, user)
+    return await get_logs(r, job_id, container_num, user, seek=seek, length=length)
 
 
 @ROUTER_JOBS.get(
@@ -378,8 +388,10 @@ async def get_job_stderr(
     job_id: _ANN_JOB_ID,
     container_num: _ANN_CONTAINER_NUMBER,
     user: CTSUser=Depends(_AUTH),
+    seek: _ANN_LOG_SEEK = None,
+    length: _ANN_LOG_LENGTH = None,
 ) -> StreamingResponse:
-    return await get_logs(r, job_id, container_num, user, stderr=True)
+    return await get_logs(r, job_id, container_num, user, stderr=True, seek=seek, length=length)
 
 
 async def get_logs(
@@ -388,9 +400,13 @@ async def get_logs(
     container_num: int,
     user: CTSUser,
     stderr: bool = False,
+    seek: int | None = None,
+    length: int | None = None,
 ) -> StreamingResponse:
     jobstate = app_state.get_app_state(r).job_state
-    filegen, filename = await jobstate.stream_job_logs(job_id, container_num, user, stderr=stderr)
+    filegen, filename = await jobstate.stream_job_logs(
+        job_id, container_num, user, stderr=stderr, seek=seek, length=length
+    )
     return StreamingResponse(
         filegen,
         media_type="text/plain; charset=utf-8",
diff --git a/cdmtaskservice/s3/client.py b/cdmtaskservice/s3/client.py
index e936c3a..5ea240d 100644
--- a/cdmtaskservice/s3/client.py
+++ b/cdmtaskservice/s3/client.py
@@ -291,25 +291,46 @@ async def head(client, buk=buk, key=key):  # bind the current value of the varia
             ))
         return ret
 
-    def stream_object(self, s3path: S3Paths) -> AsyncIterator[bytes]:
+    def stream_object(
+        self, s3path: S3Paths, seek: int | None = None, length: int | None = None
+    ) -> AsyncIterator[bytes]:
         """
         Stream an object from S3.
-        
+
         Note that the S3Paths input must contain only a single path.
+
+        s3path - the path of the object to stream.
+        seek - the byte offset in the file from which to start reading.
+        length - the number of bytes to read from the file.
         """
         if len(_not_falsy(s3path, "s3path")) > 1:
             raise ValueError("Only one path is allowed")
-        return self._stream_object_generator(s3path)
+        if seek is not None:  # TODO CODE add an optional param to _check_num
+            _check_num(seek, "seek", minimum=0)
+        if length is not None:
+            _check_num(length, "length", minimum=1)
+        return self._stream_object_generator(s3path, seek=seek, length=length)
     
-    async def _stream_object_generator(self, s3path: S3Paths) -> AsyncIterator[bytes]:
+    async def _stream_object_generator(
+        self, s3path: S3Paths, seek: int | None = None, length: int | None = None
+    ) -> AsyncIterator[bytes]:
         # internal helper returns the raw response
         # we can't use the _run_commands helper method here since the client needs to stay
         # open while the result is streamed
-        try: 
+        try:
             async with self._client() as client:
                 buk, key, path = next(s3path.split_paths(include_full_path=True))
                 async def go(client, buk=buk, key=key):
-                    return await client.get_object(Bucket=buk, Key=key)
+                    kwargs = {"Bucket": buk, "Key": key}
+                    # Build Range header if seek or length is specified
+                    if seek is not None or length is not None:
+                        start = seek if seek is not None else 0
+                        if length is not None:
+                            end = start + length - 1
+                            kwargs["Range"] = f"bytes={start}-{end}"
+                        else:
+                            kwargs["Range"] = f"bytes={start}-"
+                    return await client.get_object(**kwargs)
                 go.path = path
                 res = await self._fnc_wrapper(client, go)
                 body = res["Body"]
diff --git a/test/s3/s3_client_test.py b/test/s3/s3_client_test.py
index 48a48e1..e43d345 100644
--- a/test/s3/s3_client_test.py
+++ b/test/s3/s3_client_test.py
@@ -422,6 +422,12 @@ async def _download_objects_to_file_fail(
     assert_exception_correct(got.value, expected, print_stacktrace)
 
 
+async def _stream_and_assert(s3c, file_path, expected, seek=None, length=None):
+    asynciter = s3c.stream_object(S3Paths([file_path]), seek=seek, length=length)
+    data = b"".join([chunk async for chunk in asynciter])
+    assert data == expected
+
+
 @pytest.mark.asyncio
 async def test_stream_object(minio):
     await minio.clean()  # couldn't get this to work as a fixture
@@ -434,13 +440,17 @@ async def test_stream_object(minio):
     )
 
     s3c = await _client(minio)
-    asynciter = s3c.stream_object(S3Paths(["test-bucket/test_file"]))
-    data = b"".join([chunk async for chunk in asynciter])
-    assert data == b"imsounique"
-    
-    asynciter = s3c.stream_object(S3Paths(["test-bucket/big_test_file"]))
-    data = b"".join([chunk async for chunk in asynciter])
-    assert data == b"abcdefghij" * 600000 * 2
+    await _stream_and_assert(s3c, "test-bucket/test_file", b"imsounique")
+    await _stream_and_assert(s3c, "test-bucket/big_test_file", b"abcdefghij" * 600000 * 2)
+
+    # Test with seek only - "imsounique"[5:] = "nique"
+    await _stream_and_assert(s3c, "test-bucket/test_file", b"nique", seek=5)
+
+    # Test with length only - "imsounique"[:5] = "imsou"
+    await _stream_and_assert(s3c, "test-bucket/test_file", b"imsou", length=5)
+
+    # Test with both seek and length - "imsounique"[3:8] = "ouniq"
+    await _stream_and_assert(s3c, "test-bucket/test_file", b"ouniq", seek=3, length=5)
 
 
 @pytest.mark.asyncio
@@ -452,6 +462,22 @@ async def test_stream_object_fail_bad_input(minio):
         "Only one path is allowed"
     ))
 
+    # Test with invalid seek values
+    await _stream_object_fail(
+        s3c, S3Paths(["foo/bar"]), ValueError("seek must be >= 0"), seek=-1
+    )
+    await _stream_object_fail(
+        s3c, S3Paths(["foo/bar"]), ValueError("seek must be >= 0"), seek=-100
+    )
+
+    # Test with invalid length values
+    await _stream_object_fail(
+        s3c, S3Paths(["foo/bar"]), ValueError("length must be >= 1"), length=0
+    )
+    await _stream_object_fail(
+        s3c, S3Paths(["foo/bar"]), ValueError("length must be >= 1"), length=-1
+    )
+
 
 @pytest.mark.asyncio
 async def test_stream_object_fail_bad_connection(minio):
@@ -501,10 +527,10 @@ async def test_stream_object_fail_unauthed(minio, minio_unauthed_user):
 
 
 async def _stream_object_fail(
-    cli: S3Client, path: S3Paths, expected: Exception, print_stacktrace=False
+    cli: S3Client, path: S3Paths, expected: Exception, seek=None, length=None, print_stacktrace=False
 ):
     with pytest.raises(Exception) as got:
-        res = cli.stream_object(path)
+        res = cli.stream_object(path, seek=seek, length=length)
         [chunk async for chunk in res]
     assert_exception_correct(got.value, expected, print_stacktrace)