alchemize-ai 0.1.0__tar.gz

alchemize_ai-0.1.0/.github/workflows/ci.yml

@@ -0,0 +1,64 @@
+name: CI
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+    branches: [main]
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  lint:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Install uv
+        uses: astral-sh/setup-uv@v4
+        with:
+          version: "latest"
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Install dependencies
+        run: uv sync --group dev
+
+      - name: Lint with ruff
+        run: |
+          uv run ruff check .
+          uv run ruff format --check .
+
+  test:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        python-version: ["3.11", "3.12"]
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Install uv
+        uses: astral-sh/setup-uv@v4
+        with:
+          version: "latest"
+
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Install Rust toolchain
+        uses: dtolnay/rust-toolchain@stable
+
+      - name: Install dependencies
+        run: uv sync --group dev
+
+      - name: Run tests
+        env:
+          ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
+        run: uv run pytest tests/ -v --tb=short -x

alchemize_ai-0.1.0/.github/workflows/workflow.yml

@@ -0,0 +1,50 @@
+name: Publish to PyPI
+
+on:
+  push:
+    tags:
+      - 'v*'
+  workflow_dispatch:
+
+jobs:
+  build:
+    name: Build distribution
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: '3.12'
+
+      - name: Install build
+        run: python -m pip install --upgrade build
+
+      - name: Build wheel and sdist
+        run: python -m build
+
+      - name: Upload distributions
+        uses: actions/upload-artifact@v4
+        with:
+          name: dist
+          path: dist/
+
+  publish:
+    name: Publish to PyPI
+    needs: build
+    runs-on: ubuntu-latest
+    environment:
+      name: pypi
+      url: https://pypi.org/p/alchemize-ai
+    permissions:
+      id-token: write
+    steps:
+      - name: Download distributions
+        uses: actions/download-artifact@v4
+        with:
+          name: dist
+          path: dist/
+
+      - name: Publish to PyPI
+        uses: pypa/gh-action-pypi-publish@release/v1

alchemize_ai-0.1.0/.gitignore

@@ -0,0 +1,22 @@
+# Python
+__pycache__/
+*.pyc
+*.egg-info/
+.venv/
+.pixi/
+pixi.lock
+
+# Rust build artifacts
+target/
+Cargo.lock
+
+# Secrets
+.env
+
+# IDE
+.vscode/
+.idea/
+.ruff_cache/
+
+# OS
+.DS_Store

alchemize_ai-0.1.0/LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

alchemize_ai-0.1.0/PKG-INFO

@@ -0,0 +1,30 @@
+Metadata-Version: 2.4
+Name: alchemize-ai
+Version: 0.1.0
+Summary: Alchemize: an AI agent that transpiles between any computational framework via LLM
+Author-email: Thomas Wiecki <thomas.wiecki@pymc-labs.com>
+License-File: LICENSE
+Requires-Python: >=3.11
+Requires-Dist: anthropic>=0.52
+Requires-Dist: bridgestan>=2.7.0
+Requires-Dist: click>=8.0
+Requires-Dist: cmdstanpy>=1.3.0
+Requires-Dist: marimo>=0.20.4
+Requires-Dist: numpy
+Requires-Dist: nutpie>=0.16
+Requires-Dist: pymc>=5.28
+Requires-Dist: torch>=2.10.0
+Provides-Extra: dev
+Requires-Dist: pytest; extra == 'dev'
+Requires-Dist: ruff; extra == 'dev'
+Provides-Extra: dl
+Requires-Dist: jax>=0.4; extra == 'dl'
+Requires-Dist: jaxlib>=0.4; extra == 'dl'
+Requires-Dist: torch>=2.0; extra == 'dl'
+Provides-Extra: jax
+Requires-Dist: jax>=0.4; extra == 'jax'
+Requires-Dist: jaxlib>=0.4; extra == 'jax'
+Provides-Extra: test
+Requires-Dist: pytest; extra == 'test'
+Provides-Extra: torch
+Requires-Dist: torch>=2.0; extra == 'torch'

alchemize_ai-0.1.0/README.md

@@ -0,0 +1,263 @@
+# Transalchemy
+
+An AI agent that acts as a compiler for computational models. It transpiles between probabilistic programming languages (PyMC, Stan), deep learning frameworks (JAX, PyTorch), and compiles to optimized Rust — with numerical validation at every step.
+
+**[Read the blog post →](https://twiecki.io/blog/2026/03/10/alchemize/)**
+
+## How it works
+
+The agent doesn't translate ops mechanically. It reasons about the full computational graph and applies the optimizations a domain expert would: loop fusion, memory pre-allocation, cache-friendly access patterns.
+
+```
+Source (PyMC/Stan/JAX/PyTorch) → Extract outputs + validation points → Claude agent loop → Target (Rust/PyMC/PyTorch/JAX) → Verify
+```
+
+1. **Extract**: Read model to get parameters, transforms, logp graph, and reference values
+2. **Generate**: Claude (as an agent with tools) generates the target implementation
+3. **Verify**: Validate logp + gradients against reference values
+4. **Iterate**: If validation fails, the agent reads errors, inspects data, and fixes code autonomously
+
+The agent has four tools: `write_rust_code`, `cargo_build`, `validate_logp`, and `read_file`. It loops until the output compiles and validates correctly. Model-specific "skills" are detected automatically:
+
+- **GP (CPU)**: Linear algebra via faer (Cholesky, solves, inverses)
+- **GP (Accelerate)**: Apple Accelerate framework (AMX coprocessor) for Apple Silicon
+- **GP (CUDA)**: GPU-accelerated via cudarc + cuSOLVER for NVIDIA GPUs
+- **ZeroSumNormal**: ZeroSum transform formulas and constraint handling
+- **Stan → Rust**: Stan model extraction via BridgeStan
+- **Stan → PyMC**: Distribution mappings, idiom translation, constraint handling
+- **JAX → PyTorch**: Functional-to-stateful translation, op mapping, weight transposition
+- **PyTorch → JAX**: Stateful-to-functional translation, pure function extraction
+- **PyTorch → Rust**: Neural net to zero-dependency Rust binary, forward + gradient validation
+
+Hardware is auto-detected: CUDA → Accelerate (Apple Silicon) → CPU fallback.
+
+## Benchmarks
+
+### Compilation (Claude Sonnet 4)
+
+| Model | Params | Tool Calls | Tokens | Result |
+|---|---|---|---|---|
+| Normal | 2 | 4 | 40K | First try |
+| Linear Regression | 3 | 4 | 54K | First try |
+| Hierarchical | 12 | 8 | 153K | Fixed gradients in 1 retry |
+| GP (ExpQuad) | 3 | 11 | 467K | Passed (GP skill) |
+| ZeroSumNormal | 142 | 9 | 484K | Passed (ZeroSumNormal skill) |
+
+### Runtime: logp+dlogp evaluation speed
+
+Rust vs nutpie's Numba backend (500K evaluations, lower is better):
+
+| Model | Numba (us/eval) | Rust (us/eval) | Speedup |
+|---|---|---|---|
+| Normal (2 params) | 0.96 | 0.14 | **6.8x** |
+| LinReg (3 params) | 1.60 | 0.33 | **4.9x** |
+| Hierarchical (12 params) | 2.63 | 0.76 | **3.5x** |
+| GP regression (3 params) | 116.57 | 35.31 | **3.3x** |
+
+Numba column = `numba.cfunc` called from Rust in a tight loop (how nutpie actually works). The AI-compiled Rust is 3-7x faster across all models.
+
+### Apple Accelerate (AMX coprocessor) acceleration
+
+For GP models on Apple Silicon, the compiler auto-detects the platform and uses Apple's Accelerate framework via direct LAPACK FFI (`dpotrf`, `dpotrs`, `dpotri`). This leverages the AMX coprocessor for hardware-accelerated matrix operations in full f64 precision. Benchmarks on M4 Max, N=200 GP:
+
+| Backend | µs/eval | Speedup vs faer |
+|---|---|---|
+| Rust + faer (pure Rust) | 487 | 1.0x |
+| Rust + Accelerate (AMX) | 366 | **1.33x** |
+
+No extra crate dependencies needed — Accelerate is linked via `build.rs` to the system framework.
+
+### Neural network inference: PyTorch → Rust
+
+minGPT-nano (3 layers, 3 heads, 48-dim embeddings) transpiled to zero-dependency Rust:
+
+| Backend | µs/call | Speedup |
+|---|---|---|
+| PyTorch (eager) | 660 | 1.0x |
+| Rust (transpiled) | 284 | **2.3x** |
+| Rust + Enzyme (forward+backward) | 899 | **3.1x** vs PyTorch fwd+bwd (2777 µs) |
+
+Forward pass numerical accuracy: max_diff = 5.36e-07. Enzyme gradients match PyTorch autograd to ~5e-7 (f32 precision).
+
+## Install
+
+```bash
+pip install alchemize-ai
+```
+
+> Note: the PyPI distribution name is `alchemize-ai` (the `alchemize` name was taken by an abandoned package), but the import name is still `alchemize`:
+>
+> ```python
+> from alchemize import compile_model
+> ```
+
+The CLI command is also `alchemize`.
+
+## Quick start
+
+```bash
+export ANTHROPIC_API_KEY=sk-ant-...
+pip install alchemize-ai  # or, for development: uv sync / pip install -e .
+```
+
+```python
+import pymc as pm
+from alchemize import compile_model
+
+with pm.Model() as model:
+    mu = pm.Normal("mu", 0, 10)
+    sigma = pm.HalfNormal("sigma", 5)
+    y = pm.Normal("y", mu=mu, sigma=sigma, observed=data)
+
+result = compile_model(model)
+
+if result.success:
+    print(f"Compiled in {result.n_attempts} build(s), {result.token_usage['total_tokens']} tokens")
+```
+
+## Examples
+
+```bash
+# Single models
+python examples/01_normal.py
+python examples/02_linear_regression.py
+python examples/03_hierarchical.py
+
+# Full benchmark suite
+python examples/run_benchmark.py
+
+# logp+dlogp evaluation benchmark (Rust vs nutpie/Numba)
+python examples/bench_logp.py
+
+# Stan → PyMC transpilation
+python examples/stan_pymc_01_normal.py
+python examples/stan_pymc_02_hierarchical.py
+
+# JAX ↔ PyTorch transpilation
+python examples/jax_to_pytorch_mlp.py
+python examples/pytorch_to_jax_mlp.py
+
+# PyTorch → Rust (zero-dependency inference binary)
+python examples/pytorch_to_rust_mlp.py
+
+# minGPT → Rust (transformer inference)
+python examples/mingpt_to_rust.py
+```
+
+## Architecture
+
+```
+alchemize/
+├── exporter.py       # Extract parameters, transforms, logp graph from pm.Model()
+├── compiler.py       # Agentic loop: Claude API → Rust code → build → validate
+├── stan_exporter.py  # Extract Stan model context via BridgeStan
+├── stan_compiler.py  # Stan → Rust agentic compiler
+├── stan_to_pymc.py           # Stan → PyMC agentic transpiler
+├── jax_exporter.py           # Extract model info from JAX functions
+├── pytorch_exporter.py       # Extract model info from PyTorch modules
+├── jax_pytorch_transpiler.py  # JAX ↔ PyTorch agentic transpiler
+├── pytorch_rust_transpiler.py # PyTorch → Rust agentic transpiler (zero-dep inference)
+├── nutpie_bridge.py          # nutpie integration: compiled Rust → nutpie.sample()
+├── benchmark.py              # logp eval benchmarks: Rust vs Numba (jit + cfunc)
+└── skills/                   # Model-specific knowledge for the AI agent
+    ├── gp.md              # CPU GP (faer Cholesky)
+    ├── gp_accelerate.md   # Apple Silicon GP (Accelerate LAPACK / AMX)
+    ├── gp_cuda.md         # NVIDIA GPU GP (cudarc + cuSOLVER)
+    ├── zerosumnormal.md
+    ├── stan.md            # Stan → Rust translation knowledge
+    ├── stan_to_pymc.md    # Stan → PyMC translation knowledge
+    ├── jax_to_pytorch.md  # JAX → PyTorch op mapping + idioms
+    ├── pytorch_to_jax.md  # PyTorch → JAX op mapping + idioms
+    └── pytorch_to_rust.md # PyTorch → Rust: matmul, activations, backprop
+
+rust_template/      # Template Rust project (Cargo.toml, data loading, validation)
+bench_runner/       # Rust lib for calling Numba cfunc from Rust (like nutpie)
+compiled_models/    # Pre-compiled models (normal, linreg, hierarchical, GP, ...)
+```
+
+## Stan → PyMC Transpiler
+
+The same agentic architecture works for language-to-language translation. Claude generates PyMC code, validates logp against BridgeStan reference values, and iterates until the models match numerically. Used to translate all 120 models from [posteriordb](https://github.com/stan-dev/posteriordb) from Stan to PyMC.
+
+```python
+from alchemize import transpile_stan_to_pymc
+
+stan_code = """
+data { int<lower=0> N; array[N] real y; }
+parameters { real mu; real<lower=0> sigma; }
+model { mu ~ normal(0, 10); sigma ~ normal(0, 5); y ~ normal(mu, sigma); }
+"""
+
+result = transpile_stan_to_pymc(stan_code, data={"N": 100, "y": [...]})
+if result.success:
+    model = result.get_model(data)  # returns a pm.Model
+    print(result.pymc_code)         # generated Python code
+```
+
+## JAX ↔ PyTorch Transpiler
+
+The same agentic architecture generalizes to deep learning frameworks. Claude translates between JAX's functional style and PyTorch's stateful modules, validating forward pass outputs and gradients at multiple test points.
+
+### JAX → PyTorch
+
+```python
+import jax.numpy as jnp
+from alchemize import transpile_jax_to_pytorch
+
+def forward(params, x):
+    x = jax.nn.relu(x @ params["w1"] + params["b1"])
+    return x @ params["w2"] + params["b2"]
+
+params = {"w1": jnp.ones((4, 8)), "b1": jnp.zeros(8),
+          "w2": jnp.ones((8, 2)), "b2": jnp.zeros(2)}
+
+result = transpile_jax_to_pytorch(forward, params, sample_input=jnp.ones((1, 4)))
+if result.success:
+    model = result.get_model(params)  # returns a torch.nn.Module
+```
+
+### PyTorch → JAX
+
+```python
+import torch.nn as nn
+from alchemize import transpile_pytorch_to_jax
+
+class MLP(nn.Module):
+    def __init__(self):
+        super().__init__()
+        self.fc1 = nn.Linear(4, 8)
+        self.fc2 = nn.Linear(8, 2)
+
+    def forward(self, x):
+        return self.fc2(torch.relu(self.fc1(x)))
+
+result = transpile_pytorch_to_jax(MLP(), sample_input=torch.randn(1, 4))
+if result.success:
+    jax_params, forward_fn = result.get_model(param_data)
+    output = forward_fn(jax_params, jnp.ones((1, 4)))
+```
+
+## PyTorch → Rust Transpiler
+
+The killer feature for inference deployment. Takes a PyTorch `nn.Module` and generates a **zero-dependency Rust binary** — no ML framework, no Python runtime, just raw f32 math compiled to native code. Parameters are baked in as `const` arrays.
+
+The agent uses the same agentic architecture with tools: `write_code` → `cargo_build` → `validate_model` (forward pass + gradient matching). The generated Rust code includes both `forward()` and manual backpropagation for gradient validation.
+
+```python
+import torch.nn as nn
+from alchemize import transpile_pytorch_to_rust
+
+class MLP(nn.Module):
+    def __init__(self):
+        super().__init__()
+        self.fc1 = nn.Linear(4, 8)
+        self.fc2 = nn.Linear(8, 2)
+
+    def forward(self, x):
+        return self.fc2(torch.relu(self.fc1(x)))
+
+result = transpile_pytorch_to_rust(MLP(), sample_input=torch.randn(1, 4))
+if result.success:
+    print(f"Binary at: {result.binary_path}")  # Zero-dependency native binary
+    result.save("model.rs")                     # Save the generated Rust code
+```