diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
index da45237b4..5c112c44c 100644
--- a/.github/workflows/build.yml
+++ b/.github/workflows/build.yml
@@ -128,7 +128,7 @@ jobs:
           repository: laminlabs/laminapp-ui
           token: ${{ secrets.GH_TOKEN_DEPLOY_LAMINAPP }}
           path: laminhub
-          ref: setup
+          ref: main
       - uses: actions/setup-python@v4
         with:
           python-version: '3.10'
diff --git a/lamindb_setup/core/hashing.py b/lamindb_setup/core/hashing.py
new file mode 100644
index 000000000..0819252be
--- /dev/null
+++ b/lamindb_setup/core/hashing.py
@@ -0,0 +1,62 @@
+"""Hashing.
+
+.. autosummary::
+   :toctree: .
+
+   hash_set
+   hash_file
+
+"""
+
+import base64
+import hashlib
+from typing import List, Set, Tuple
+
+
+def to_b64_str(bstr: bytes):
+    b64 = base64.urlsafe_b64encode(bstr).decode().strip("=")
+    return b64
+
+
+def b16_to_b64(s: str):
+    return to_b64_str(base64.b16decode(s.strip('"'), casefold=True))
+
+
+# a lot to read about this: lamin-notes/2022/hashing
+def hash_set(s: Set[str]) -> str:
+    bstr = ":".join(sorted(s)).encode("utf-8")
+    # as we're truncating at 20 b64, we choose md5 over sha512
+    return to_b64_str(hashlib.md5(bstr).digest())[:20]
+
+
+def hash_md5s_from_dir(etags: List[str]) -> Tuple[str, str]:
+    # need to sort below because we don't want the order of parsing the dir to
+    # affect the hash
+    digests = b"".join(
+        hashlib.md5(etag.encode("utf-8")).digest() for etag in sorted(etags)
+    )
+    digest = hashlib.md5(digests).digest()
+    return to_b64_str(digest)[:22], "md5-d"
+
+
+def hash_file(file_path, chunk_size=50 * 1024 * 1024) -> Tuple[str, str]:
+    chunks = []
+    with open(file_path, "rb") as fp:
+        # read first chunk
+        chunks = [fp.read(chunk_size)]
+        # try reading the 2nd chunk
+        data = fp.read(chunk_size)
+        if data:
+            # go to end of file
+            fp.seek(-chunk_size, 2)
+            # read last chunk
+            data = fp.read(chunk_size)
+            chunks.append(data)
+    if len(chunks) == 1:
+        digest = hashlib.md5(chunks[0]).digest()
+        hash_type = "md5"
+    else:
+        digests = b"".join(hashlib.sha1(chunk).digest() for chunk in chunks)
+        digest = hashlib.sha1(digests).digest()
+        hash_type = "sha1-fl"  # sha1 first last chunk
+    return to_b64_str(digest)[:22], hash_type
diff --git a/lamindb_setup/core/upath.py b/lamindb_setup/core/upath.py
index ae4b371de..0d0441cfb 100644
--- a/lamindb_setup/core/upath.py
+++ b/lamindb_setup/core/upath.py
@@ -9,7 +9,7 @@
 from typing import Literal, Dict
 import fsspec
 from itertools import islice
-from typing import Optional, Set, Any
+from typing import Optional, Set, Any, Tuple
 from collections import defaultdict
 from dateutil.parser import isoparse  # type: ignore
 from lamin_utils import logger
@@ -17,6 +17,7 @@
 from upath.implementations.cloud import CloudPath, S3Path  # noqa  # keep CloudPath!
 from upath.implementations.local import LocalPath, PosixUPath, WindowsUPath
 from .types import UPathStr
+from .hashing import b16_to_b64, hash_md5s_from_dir
 
 LocalPathClasses = (PosixUPath, WindowsUPath, LocalPath)
 
@@ -484,3 +485,65 @@ def create_path(path: UPath, access_token: Optional[str] = None) -> UPath:
         secret=credentials["secret"],
         token=credentials["token"],
     )
+
+
+def get_stat_file_cloud(stat: Dict) -> Tuple[int, str, str]:
+    size = stat["size"]
+    # small files
+    if "-" not in stat["ETag"]:
+        # only store hash for non-multipart uploads
+        # we can't rapidly validate multi-part uploaded files client-side
+        # we can add more logic later down-the-road
+        hash = b16_to_b64(stat["ETag"])
+        hash_type = "md5"
+    else:
+        stripped_etag, suffix = stat["ETag"].split("-")
+        suffix = suffix.strip('"')
+        hash = f"{b16_to_b64(stripped_etag)}-{suffix}"
+        hash_type = "md5-n"  # this is the S3 chunk-hashing strategy
+    return size, hash, hash_type
+
+
+def get_stat_dir_s3(path: UPath) -> Tuple[int, str, str, int]:
+    import boto3
+
+    if not AWS_CREDENTIALS_PRESENT:
+        # passing the following param directly to Session() doesn't
+        # work, unfortunately: botocore_session=path.fs.session
+        from botocore import UNSIGNED
+        from botocore.config import Config
+
+        config = Config(signature_version=UNSIGNED)
+        s3 = boto3.session.Session().resource("s3", config=config)
+    else:
+        s3 = boto3.session.Session().resource("s3")
+    bucket, key, _ = path.fs.split_path(path.as_posix())
+    # assuming this here is the fastest way of querying for many objects
+    objects = s3.Bucket(bucket).objects.filter(Prefix=key)
+    size = sum([object.size for object in objects])
+    md5s = [
+        # skip leading and trailing quotes
+        object.e_tag[1:-1]
+        for object in objects
+    ]
+    n_objects = len(md5s)
+    hash, hash_type = hash_md5s_from_dir(md5s)
+    return size, hash, hash_type, n_objects
+
+
+def get_stat_dir_gs(path: UPath) -> Tuple[int, str, str, int]:
+    import google.cloud.storage as gc_storage
+
+    bucket, key, _ = path.fs.split_path(path.as_posix())
+    # assuming this here is the fastest way of querying for many objects
+    client = gc_storage.Client(
+        credentials=path.fs.credentials.credentials, project=path.fs.project
+    )
+    objects = client.Bucket(bucket).list_blobs(prefix=key)
+    sizes, md5s = [], []
+    for object in objects:
+        sizes.append(object.size)
+        md5s.append(object.md5_hash)
+    n_objects = len(md5s)
+    hash, hash_type = hash_md5s_from_dir(md5s)
+    return sum(sizes), hash, hash_type, n_objects
diff --git a/tests/test_hashing.py b/tests/test_hashing.py
new file mode 100644
index 000000000..d2688ce6b
--- /dev/null
+++ b/tests/test_hashing.py
@@ -0,0 +1,35 @@
+import base64
+from pathlib import Path
+
+from lamindb_setup.core.hashing import b16_to_b64, hash_file, to_b64_str
+
+
+def test_compute_hash():
+    files = [
+        # file_content, hash, chunk_size, hash_type
+        ("a", "DMF1ucDxtqgxw5niaXcmYQ", None, "md5"),
+        ("b", "kutf_uauL-w61xx3dTFXjw", None, "md5"),
+        ("abc", "kAFQmDzST7DWlj99KOF_cg", None, "md5"),
+        ("a", "DMF1ucDxtqgxw5niaXcmYQ", 1, "md5"),
+        ("b", "kutf_uauL-w61xx3dTFXjw", 1, "md5"),
+        # the last case here triggers multi-chunk compute with sha1
+        ("abc", "p0EbDbQEP1wS-Tw6TuBjKS", 1, "sha1-fl"),
+    ]
+    for content, hash, chunk_size, hash_type in files:
+        filepath = Path("file_1")
+        filepath.write_text(content)
+        computed_hash, computed_hash_type = hash_file(filepath, chunk_size=chunk_size)
+        assert computed_hash == hash
+        assert computed_hash_type == hash_type
+        filepath.unlink()
+
+
+def test_base64():
+    mytest = b"test"
+    b64_str = to_b64_str(mytest)
+    b64_str_padded = f"{b64_str}=="
+    assert base64.urlsafe_b64decode(b64_str_padded.encode()).hex() == mytest.hex()
+
+
+def test_b16_to_b64():
+    assert b16_to_b64("9b89c8c1acf79dba5b5341d1fff9806f") == "m4nIwaz3nbpbU0HR__mAbw"