Medical AI, built the way a
regulated device actually ships.
I design and validate deep-learning systems for clinical imaging, then build the FDA/ISO tooling — traceability matrices, audit trails, CAPA workflows — that gets them through design controls. QA automation and public-health data products round out the stack.
- Prepared by
- LSaiko
- Disciplines
- 4
- Repos under CM
- 31
- Primary language
- Python
- Effective date
- —
The two projects that best represent the two ends of the portfolio: a clinical model with built-in explainability, and a live public-facing data product.
Multi-label chest X-ray classifier over the 15-class NIH ChestX-ray14 dataset. ViT-B/16 backbone, CLAHE preprocessing, Grad-CAM heatmaps on every prediction so a radiologist has something to check the model's work against.
Real-time U.S. disease surveillance dashboard on live CDC NNDSS data — 22 notifiable diseases, all 50 states, trend calculation, and daily AI situation reports. Free, no login required.
Classification, segmentation, and tracking across radiology, dermatology, endoscopy, cardiology, and surgery — each paired with explainability rather than a black-box score.
Tooling that targets the standards real medical-device QMS teams answer to — 21 CFR 820, ISO 13485, ISO 14971, IEC 62304, 21 CFR Part 11 — turning compliance artifacts into generated, repeatable outputs.
Live, source-attributed data products and NLP tooling — built to be free, transparent, and usable without an account.
Selenium and pytest frameworks covering UI, API, cross-browser, and database-integrity testing, each wired into a CI pipeline so every suite runs the same way on every push.
The portfolio sits at an unusual intersection: the deep-learning skill to build clinical models, the regulatory fluency to push them through design controls, and the test-engineering discipline to keep all of it verifiably working.
Clinical imaging AI
Classification, segmentation, and tracking across radiology, dermatology, endoscopy, and surgery — consistently paired with explainability (Grad-CAM, bounding boxes, confidence scores) rather than black-box outputs.
Regulatory & quality
Tooling that targets the standards real medical-device teams answer to: 21 CFR 820, ISO 13485, ISO 14971, IEC 62304, 21 CFR Part 11, AIAG MSA and AIAG-VDA pFMEA.
Data & public health
Live, source-attributed data products — most prominently a CDC-backed disease surveillance dashboard with AI summaries — built to be free and usable without an account.
QA & test automation
Selenium/pytest frameworks covering UI, API, cross-browser, and database-integrity testing, with CI pipelines so every suite runs the same way on every push.