AI Product Engineer · Python · FastAPI · RAG · Computer Vision
Retrieval, computer vision, role-based workflows, tests, CI. I came from the product and funding side — ~$76K in non-dilutive funding across 5 projects, listed author on registered MedAI software — which is why I build with provenance, review workflows, and audit trails.
AI system turning operator requests, enterprise documents, public standards, and video context into review-ready industrial work-instruction drafts under mandatory expert approval. FastAPI, hybrid retrieval, video analysis, 6 roles, versioning, audit trail. Working paper on SSRN, Abstract ID 7026918.
View repositorySanitized product and architecture notes based on a private Flutter / FlutterFlow prototype.
View repositoryLimited-code case study for recognizing text on technical drawings: preprocessing, region handling, and a CRNN + CTC recognizer for narrow technical text lines.
View repositoryWorking on PDF text extraction and vector tooling in libraries I use in production.
Found and reported a class of PDFs where ff, fi and fl are dropped from extracted text. The correct text sits in /ActualText on marked content, which the library ignores. Verified the cause by patching the key in place and watching poppler's output break on exactly the affected lines.
Contributed a reproducible analysis: which sources of truth exist in the file, why the font offers no fallback, and a decoding bug in a proposed fix that doubled the corrupted output instead of removing it.
View issueTwo pull requests: a docstring describing an index-build process the code no longer performs, and an install requirement stating a Python version the package rejects.
View pull requestsProduct ownership and hands-on engineering, in medtech, industrial AI, and early-stage digital products.
I care about clear product context, simple architecture, fast validation, and tools that survive real operational use.