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. 69 merged PRs across 22 upstream repositories — verified on 23 October 2026.
Recent merges: LeRobot #4835 · REA #4414 · PyPose #437.
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 | 23,588 | #10 | 17 merged | Face-box clipping, normalization and vector-search correctness. #1655 |
| Denoising Diffusion | 10,712 | #9 | 1 merged | Preserve external conditioning and self-conditioning in 1D diffusion. #373 |
| Atomic Agents | 6,291 | #28 | 1 merged | Keep existing chat history intact when loading fails. #292 |
| Robotics Toolbox | 3,551 | #6 | 11 merged | Kinematics, dynamics, inverse kinematics and URDF correctness. #801 |
| RSL-RL | 3,086 | #6 | 2 merged | Resume RND schedules and preserve reward-history ownership. #240 |
| PyPose | 1,616 | #15 | 4 merged | Least-squares rigid alignment, robust-kernel stability and camera broadcasting. #437 |
| Geoopt | 1,100 | #3 | 4 merged | Manifold constraints, gradient projection and stable geometry. #267 |
| skrl | 1,099 | #3 | 3 merged | Gaussian policy math and cross-backend exploration consistency. #475 |
| pytransform3d | 707 | #15 | 2 merged | Stable Lie-group Jacobians and screw-axis translation handling. #389 |
| RoMa | 646 | #2 | 4 merged | Quaternion broadcasting, affine shapes and reproducible sampling. #33 |
| Rectified Flow | 489 | #3 | 1 merged | Keep NanoFlow inversion in normalized flow coordinates. #50 |
| KinPy | 181 | #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 | 368,754 | 2 merged | Deduplicate setup clients and complete positional ZIP reads. #4414 |
| LeRobot | 28,418 | 1 merged | Correct pairwise RoPE frequencies in VLA-JEPA's training-only predictor. #4835 |
| Haystack | 26,816 | 1 merged | Keep pipeline snapshots reusable after a resumed run. #13076 |
| Recommenders | 21,959 | 2 merged | Preserve custom columns and reject invalid top-k values. #2389 |
| Sentence Transformers | 19,212 | 3 merged | Ranking gradients, encoder masks and float16 quantization. #4127 |
| Kornia | 11,443 | 2 merged | Finite geometry gradients and stable small-root quartics. #5517 |
| MLxtend | 5,189 | 1 merged | Separate extracted emoticons from the preceding final word. #1207 |
| Geomstats | 1,527 | 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 69 PRs, commit counts and verification method.
TD-MPC2 (986 ★, verified contributor-list position #3): preserve target-Q history and evaluation mode during model conversion. The implementation from #98 was cherry-picked with Afloat16 authorship; the original PR was closed without merging. Together with the 22 repositories above, this makes 23 upstream projects with accepted contributions.
Robotics Toolbox also incorporated the circular particle-heading work from #735 into #883, retaining co-authorship. These adoptions are recorded separately from the 69 merged author PRs. Details.
Anti-REA Forge — MIT · v0.4. Native release hardening with CMake-ready build integration, PE/ELF checks, guarded C++ memory and one-use request authorization. Explore runnable examples, cross-platform CI and research-guided design. Project story · Join the community.
| 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.