mehta-lab
diff --git a/‎.github/workflows/test.yml‎
Lines changed: 52 additions & 1 deletion b/‎.github/workflows/test.yml‎
Lines changed: 52 additions & 1 deletion
diff --git a/‎.planning/MILESTONES.md‎
Lines changed: 15 additions & 0 deletions b/‎.planning/MILESTONES.md‎
Lines changed: 15 additions & 0 deletions
diff --git a/‎.planning/PROJECT.md‎
Lines changed: 45 additions & 58 deletions b/‎.planning/PROJECT.md‎
Lines changed: 45 additions & 58 deletions
@@ -40,10 +40,61 @@ jobs:
         run: uv run --frozen pytest --cov=src/ --cov-report=term-missing
         working-directory: packages/${{ matrix.package }}
 
+  test-data:
+    name: Test Data (Python ${{ matrix.python-version }}, ${{ matrix.os }})
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: true
+      matrix:
+        os: [ubuntu-latest, macos-latest, windows-latest]
+        python-version: ["3.11", "3.12", "3.13"]
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v5
+
+      - name: Set up uv with Python ${{ matrix.python-version }}
+        uses: astral-sh/setup-uv@v7
+        with:
+          python-version: ${{ matrix.python-version }}
+          enable-cache: true
+          cache-suffix: ${{ matrix.os }}-${{ matrix.python-version }}
+
+      - name: Install dependencies
+        run: uv sync --frozen --dev
+        working-directory: packages/viscy-data
+
+      - name: Run tests with coverage
+        run: uv run --frozen pytest --cov=viscy_data --cov-report=term-missing
+        working-directory: packages/viscy-data
+
+  test-data-extras:
+    name: Test Data Extras (Python 3.13, ubuntu-latest)
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v5
+
+      - name: Set up uv with Python 3.13
+        uses: astral-sh/setup-uv@v7
+        with:
+          python-version: "3.13"
+          enable-cache: true
+          cache-suffix: ubuntu-latest-3.13
+
+      - name: Install dependencies
+        run: uv sync --frozen --all-extras --dev
+        working-directory: packages/viscy-data
+
+      - name: Run tests with coverage
+        run: uv run --frozen pytest --cov=viscy_data --cov-report=term-missing
+        working-directory: packages/viscy-data
+
   check:
     name: All tests pass
     if: always()
-    needs: [test]
+    needs: [test, test-data, test-data-extras]
     runs-on: ubuntu-latest
     steps:
       - name: Verify all test jobs succeeded
 
@@ -1,5 +1,20 @@
 # Milestones: VisCy Modularization
 
+## v1.1 — Extract viscy-data
+
+**Shipped:** 2026-02-14
+**Phases:** 6–9 (4 executed)
+
+**What shipped:**
+- `viscy-data` package extracted with src layout (15 modules, 4015 LOC)
+- 45 public exports with clean import paths: `from viscy_data import X`
+- Optional dependency groups: `[triplet]`, `[livecell]`, `[mmap]`, `[all]`
+- Shared utilities extracted from hcs.py into _utils.py
+- All existing data tests passing (71 tests)
+- Tiered CI for viscy-data (3x3 base + 1x1 extras)
+
+---
+
 ## v1.0 — Transforms & Monorepo Skeleton
 
 **Shipped:** 2026-01-29
 
@@ -2,83 +2,67 @@
 
 ## What This Is
 
-Restructuring VisCy from a monolithic package into a uv workspace monorepo. This enables reusing transforms, dataloaders, and models in downstream projects without requiring the entire VisCy package as a dependency. The second milestone extracts `viscy-models` as an independent package containing all 8 network architectures as pure nn.Modules, organized by function (unet/, vae/, contrastive/) with shared components factored into a `components/` module.
+Restructuring VisCy from a monolithic package into a uv workspace monorepo. This enables reusing transforms, dataloaders, and models in downstream projects without requiring the entire VisCy package as a dependency. Two subpackages are now extracted: `viscy-transforms` (v1.0) and `viscy-data` (v1.1).
 
 ## Core Value
 
-**Independent, reusable subpackages with clean import paths.** Users can `pip install viscy-models` and use `from viscy_models import UNeXt2` without pulling in the entire VisCy ecosystem.
-
-## Current Milestone: v1.1 Models
-
-**Goal:** Extract all network architectures into `viscy-models` as pure nn.Modules with comprehensive test coverage.
-
-**Target features:**
-- `viscy-models` package with 8 architectures organized by function (unet/, vae/, contrastive/)
-- Shared components extracted to `components/` (stems, heads, decoder blocks)
-- Full test coverage: migrate existing + write new for UNeXt2, ContrastiveEncoder, BetaVAE
-- Independent of viscy-transforms and lightning (torch/timm/monai deps only)
-- State dict key compatibility preserved for checkpoint loading
+**Independent, reusable subpackages with clean import paths.** Users can `pip install viscy-transforms` or `pip install viscy-data` and use clean imports without pulling in the entire VisCy ecosystem.
 
 ## Requirements
 
 ### Validated
 
-- ✓ uv workspace scaffolding with `packages/` directory structure — v1.0
-- ✓ `viscy-transforms` package extracted with src layout — v1.0
-- ✓ Import path: `from viscy_transforms import X` (clean break) — v1.0
-- ✓ hatchling build backend with uv-dynamic-versioning — v1.0
-- ✓ All existing transform tests passing in new structure — v1.0
-- ✓ CI updated for monorepo structure — v1.0
+- uv workspace scaffolding with `packages/` directory structure — v1.0
+- `viscy-transforms` package extracted with src layout — v1.0
+- Import path: `from viscy_transforms import X` (clean break) — v1.0
+- hatchling build backend with uv-dynamic-versioning — v1.0
+- All existing transform tests passing in new structure — v1.0
+- CI for monorepo (9-job test matrix + lint) — v1.0
+- `viscy-data` package extracted with src layout (15 modules, 4015 LOC) — v1.1
+- All 13 data modules migrated with clean import paths — v1.1
+- Import path: `from viscy_data import X` (45 public exports) — v1.1
+- No dependency on viscy-transforms (BatchedCenterSpatialCropd in _utils.py) — v1.1
+- Optional dependency groups: `[triplet]`, `[livecell]`, `[mmap]`, `[all]` — v1.1
+- Shared utilities extracted from hcs.py into _utils.py — v1.1
+- All existing data tests passing (71 tests) — v1.1
+- Tiered CI for viscy-data (3x3 base + 1x1 extras) — v1.1
 
 ### Active
 
-- [ ] `viscy-models` package with src layout and function-based organization
-- [ ] All 8 architectures migrated: UNeXt2, FCMAE, ContrastiveEncoder, ResNet3dEncoder, BetaVae25D, BetaVaeMonai, Unet2d, Unet25d
-- [ ] Shared components extracted to `components/` (stems, heads, blocks)
-- [ ] Full test coverage: existing tests migrated + new tests for untested models
-- [ ] Import path: `from viscy_models import UNeXt2` (clean break)
-- [ ] State dict key compatibility preserved
-- [ ] CI includes viscy-models in test matrix
+(None — next milestone not yet defined)
 
 ### Out of Scope
 
-- Hydra/BaseModel registry infrastructure — deferred to future `viscy-hydra` package
-- BaseModelMeta metaclass — deferred to future `viscy-hydra` package
-- Extracting data packages (viscy-data) — future milestone
-- Application-level LightningModules — move to applications/ in future milestone
+- Extracting viscy-models, viscy-airtable — future milestones
+- Meta-package with re-exports — decided against, clean break approach
 - Backward-compatible imports — not maintaining
-- Documentation (Zensical + GitHub Pages) — deferred from v1.0
+- Zensical documentation / GitHub Pages — deferred
+- Hydra integration — per design doc
+- GPU transform unification (GPUTransformMixin) — future refactor
 
 ## Context
 
 **Design doc:** https://github.com/mehta-lab/VisCy/issues/353
 
-**Reference implementations:**
-- flowbench BaseModel: `/home/eduardo.hirata/repos/flowbench/src/models/base_model.py` (future reference for viscy-hydra)
-- lightning-hydra-template: https://github.com/ashleve/lightning-hydra-template (future reference)
-- iohub pyproject.toml: modern hatchling + uv-dynamic-versioning pattern
-
-**Current state (post v1.0):**
-- uv workspace with `viscy-transforms` at `packages/viscy-transforms/`
-- Root `viscy` umbrella package with dynamic versioning
-- Models in monolithic `viscy/unet/networks/` and `viscy/representation/`
-- 14+ shared components in unext2.py used by fcmae, contrastive, vae
-- No tests for UNeXt2, ContrastiveEncoder, or BetaVAE
+**Current state (after v1.1):**
+- uv workspace monorepo with 2 extracted packages:
+  - `packages/viscy-transforms/` — 16 transform modules, 44 exports
+  - `packages/viscy-data/` — 15 data modules, 45 exports, 4015 LOC source + 671 LOC tests
+- CI: test.yml (viscy-transforms 3x3, viscy-data 3x3 + extras 1x1) + lint.yml
+- Python >=3.11, hatchling + uv-dynamic-versioning
+- Original code on `main` branch for reference
 
-**Architecture vision:**
-- `viscy-models`: Pure nn.Module architectures (this milestone)
-- `viscy-hydra`: BaseModel, BaseModelMeta, Hydra config utilities (future milestone)
-- Applications (Cytoland, DynaCLR): LightningModules composing models with training logic (future milestone)
+**Architecture reference:**
+- `viscy/data/README.md` documents module inventory, class hierarchy, training pipeline mapping, GPU transform patterns
 
 ## Constraints
 
-- **Package naming**: `viscy-models` (hyphen) as package name, `viscy_models` (underscore) as import
-- **Python version**: >=3.11 (matching current VisCy)
-- **Build system**: hatchling with uv-dynamic-versioning (following viscy-transforms pattern)
-- **Layout**: src layout required (`packages/viscy-models/src/viscy_models/`)
-- **Independence**: viscy-models must NOT depend on viscy-transforms or lightning
-- **Dependencies**: torch, timm, monai, numpy only
-- **Tooling**: uv only, no pip/setuptools for package management
+- **Package naming**: hyphen for package name, underscore for import
+- **Python version**: >=3.11
+- **Build system**: hatchling with uv-dynamic-versioning
+- **Layout**: src layout (`packages/*/src/*/`)
+- **Tooling**: uv only
+- **No cross-package dependencies between data and transforms**
 
 ## Key Decisions
 
@@ -88,10 +72,13 @@ Restructuring VisCy from a monolithic package into a uv workspace monorepo. This
 | hatchling over setuptools | Modern, faster, better uv integration | ✓ Good |
 | src layout | Prevents import confusion during development | ✓ Good |
 | Tests inside packages | Isolated testing, `uv run --package` workflow | ✓ Good |
-| Pure nn.Module in viscy-models | No Lightning/Hydra coupling; maximum reusability | — Pending |
-| Hydra infra in separate package | Keeps model package lightweight; Hydra optional for consumers | — Pending |
-| Function-based grouping (unet/, vae/, contrastive/) | Clean organization for 8+ models with shared components | — Pending |
-| viscy-models independent of viscy-transforms | Keep packages loosely coupled | — Pending |
+| No viscy-transforms dep in data | Transforms separate from data | ✓ Good |
+| Optional dependency groups | Heavy deps as extras, lean base install | ✓ Good |
+| Extract shared utils from hcs.py | Prevent dual-role module anti-pattern | ✓ Good |
+| BatchedCenterSpatialCropd in _utils.py | CenterSpatialCropd can't handle batch dim in on_after_batch_transfer | ✓ Good |
+| Lazy imports for optional deps | try/except at module level, guard in __init__ | ✓ Good |
+| Flat public API (45 exports) | MONAI pattern, consistent with viscy-transforms | ✓ Good |
+| combined.py preserved as-is | No split per REF-02 deferral | ✓ Good |
 
 ---
-*Last updated: 2026-02-12 after requirements definition*
+*Last updated: 2026-02-14 after v1.1 milestone completion*