My research focuses on robotics, large language models, and 3D reconstruction.
I build reproducible research tools, explore native software protection, and contribute focused upstream fixes.
Python · TypeScript · C/C++ · Reproducible experiments · Upstream engineering
I contribute fixes in robotics, machine learning, geometry and developer tools. 66 merged PRs across 21 upstream repositories — verified on 19 October 2026.
Verified top-30 contributor-list positions come first; projects within each group are ordered by upstream stars. Every count links to my merged PRs in that repository.
| Project | Upstream ★ | Position¹ | Merged PRs | Representative work |
|---|---|---|---|---|
| DeepFace | 23,571 | #10 | 17 merged | Face-box clipping, normalization and vector-search correctness. #1655 |
| Denoising Diffusion | 10,704 | #9 | 1 merged | Preserve external conditioning and self-conditioning in 1D diffusion. #373 |
| Atomic Agents | 6,287 | #28 | 1 merged | Keep existing chat history intact when loading fails. #292 |
| Robotics Toolbox | 3,546 | #6 | 11 merged | Kinematics, dynamics, inverse kinematics and URDF correctness. #801 |
| RSL-RL | 3,074 | #6 | 2 merged | Resume RND schedules and preserve reward-history ownership. #240 |
| PyPose | 1,616 | #16 | 3 merged | Robust-kernel numerical stability and camera-axis broadcasting. #417 |
| skrl | 1,101 | #3 | 3 merged | Gaussian policy math and cross-backend exploration consistency. #475 |
| Geoopt | 1,100 | #3 | 4 merged | Manifold constraints, gradient projection and stable geometry. #267 |
| pytransform3d | 707 | #15 | 2 merged | Stable Lie-group Jacobians and screw-axis translation handling. #389 |
| RoMa | 645 | #2 | 4 merged | Quaternion broadcasting, affine shapes and reproducible sampling. #33 |
| Rectified Flow | 488 | #3 | 1 merged | Keep NanoFlow inversion in normalized flow coordinates. #50 |
| KinPy | 182 | #2 | 3 merged | World-frame Jacobians with consistent axes and link offsets. #29 |
| Elements | 73 | #3 | 2 merged | Preserve hidden filenames and correct local-path operations. #19 |
| qpmpc | 55 | #2 | 1 merged | Preserve absent state constraints in receding-horizon updates. #15 |
| Project | Upstream ★ | Merged PRs | Representative work |
|---|---|---|---|
| REA | 331,627 | 1 merged | Deduplicate selected clients before generating setup plans. #3919 |
| Haystack | 26,773 | 1 merged | Keep pipeline snapshots reusable after a resumed run. #13076 |
| Recommenders | 21,957 | 2 merged | Preserve custom columns and reject invalid top-k values. #2389 |
| Sentence Transformers | 19,192 | 3 merged | Ranking gradients, encoder masks and float16 quantization. #4127 |
| Kornia | 11,435 | 2 merged | Finite geometry gradients and stable small-root quartics. #5517 |
| MLxtend | 5,187 | 1 merged | Separate extracted emoticons from the preceding final word. #1207 |
| Geomstats | 1,525 | 1 merged | Propagate product-metric parallel transport to each factor. #2131 |
¹ Positions follow GitHub’s commit-sorted contributor list, not PR counts or maintainer status. Equal commit counts follow GitHub’s returned order. Stars and positions are dated snapshots. The second table includes repositories without a verified top-30 placement. All 66 PRs, commit counts and verification method.
Anti-REA Forge — Open research on native software protection against AI reverse engineering. CMake/C++, PE/ELF release gates, and reproducible experiments. Contributions welcome.
| Robotics | Large language models | 3D reconstruction |
|---|---|---|
| Bag Timeline Audit Read-only ROS 2 SQLite/MCAP timestamp diagnostics. |
LLM Handshake Bounded OpenAI-compatible endpoint probes and report comparisons. |
ReconTrail CPU photogrammetry, sparse geometry and offline reports. |
| TF Route Audit Exported TF and native URDF topology/time-window checks. |
SSE Toolcall Audit and Tooltrace Audit Saved stream and tool-execution diagnostics. |
Pose Convention Bridge and COLMAP Model Audit Explicit pose conversion and text-model preflight. |
Research and engineering workflow: Research to Fix · Debate Direction · StewardCheck · Agent Action Notifier.
How I work: isolate the symptom, understand the boundary, write a regression, and make the result inspectable. The six domain utilities have 67 local tests and 36 passing OS/Python CI combinations at the recorded release. These are experimental diagnostics and conversion tools; each repository documents its limits.