Machine Learning Engineer & Technical Lead | LLM/RAG & Document Intelligence | Scientific ML & Medical Imaging
I build applied AI and research software for complex documents and scientific data. My work spans retrieval and agent workflows, evaluation, backend engineering, and reproducible scientific ML.
Start with the public projects and runnable examples below. The CV provides fuller professional context.
| Project | Role in the portfolio | Verifiable evidence |
|---|---|---|
| Yandex Workspace MCP | Open-source integration engineering | Offline permission demo, typed tools, permission gates, audit logging, CI, Docker and deployment docs |
| mcp-capguard | MCP tooling and permission testing | pytest plugin + CLI for checking exposed tools; runnable FastMCP regression example |
| Tensor-Based Modal Decomposition | Scientific ML and research software | Tucker/HOSVD, sparse sensing, tests, synthetic run, reproducibility guide, preprint |
| LLM Adviser case study | Architecture and technical leadership | Sanitized account of document processing, retrieval, evaluation, and reliability decisions; application source is private |
For a smaller document-processing example, see the OCR + LLM pipeline. Its synthetic fixture demonstrates the workflow; real-document accuracy is not established there.
The projects show different parts of the work: MCP integration, reusable testing tools, scientific methods, and system design.
Merged upstream:
- dragonflydb/dragonfly #8199 — fixed
FT.INFOmissing-index wording for RedisVL compatibility, with a regression test; merged 1 September 2026. - wandb/rai-toolkit #24 — added an HR industry preset, dataset selection, documentation, and focused tests; merged 2 September 2026.
- IBM/docling-pipelines #31 — made example-test dry runs offline and side-effect free by returning before prerequisite probes; added regression and control tests; merged 9 September 2026.
Additional PRs: PageIndex plain-text indexing and Weaviate generative integration. Their PR pages show current status and scope.
Tools and methods used across professional, research, and open-source work. The projects above provide public examples of selected areas.
- Machine Learning Engineer / Technical Lead, Analytics and Machine Learning Department, MSUU (
Apr 2024 – Present) — architecture and delivery across document and retrieval systems; engineering leadership and evaluation practices. - Data Scientist / Research ML Engineer, Heriot-Watt TPU Center (
Sep 2024 – Present) — Bayesian parameter-space appraisal, tensor-based reduced-order modeling, and QR sparse-sensor placement. - Machine Learning Engineer, Cardiology Research Institute (
Mar 2021 – May 2025) — EPIFAT cardiac CT segmentation software and medical-imaging research workflows.
Full role scope and employment details: CV · LinkedIn. EPIFAT is state-registered software (certificate No. 2025610317); related medical code and data are not published here.
- arXiv:2607.09687 — Tensor-Based Modal Decomposition and Sparse Sensor Placement for the Brugge Field Simulation Model.
- Digital Diagnostics — cardiac MRI radiomics research.
- ORCID — complete research identity and publication record.
- Google Scholar — publication and citation profile.
Open to research and open-source collaboration. For project-specific questions, please use the relevant repository's issues or discussions.