I do computer vision, LLM pipelines, edge AI, and NLP. Lately I’ve been doing weld detection on fuel pipelines, tracking aircraft from a Raspberry Pi, and building a local OCR app that runs entirely on-device. I also build my own products on the side.
Full CV pipeline for detecting welds in pipeline inspection data. Built it from scratch, then found a position-dependent sensor bias hiding in the data that nobody had noticed before. This runs on critical infrastructure — fuel pipelines across the country.
Raspberry Pi + Hailo NPU that detects aircraft, locks a gimbal onto them, measures distance with a laser, fuses ADS-B transponder data, and calculates GPS coordinates. All real-time, all on edge.
30+ page platform with ABAC policy engine, hash-chain audit trails, and mindmap document views. Built the compliance architecture from scratch because the off-the-shelf options couldn’t handle classification levels properly.
Person extraction from 128 Polish regulatory PDFs. Tested 46 LLMs through a unified evaluation pipeline, then wrote preprocessing that shrank each file enough for near-free inference. The hard part was picking the right model, not running it.
On-device OCR that runs a vision-language model entirely on your Mac. Tables, formulas, charts, seals. Fully offline — your data never leaves the machine.
Independent benchmark tracking across 17 ML research areas with 286+ verified results and implementation links.
Conversational recommendation engine that builds a taste profile through dialogue — context-aware picks, not just collaborative filtering.
RAG solutions for enterprise clients. LLM pipeline monitoring with Langfuse. Model selection and retrieval optimization.
Signal and image processing for industrial non-destructive testing. Anomaly detection, binary sensor data pipelines, visualization dashboards.
GenAI platform integrating LLMs into business workflows. FastAPI, LangChain, LlamaIndex. Production deployments.
Transaction systems, API integrations, async processing for a fintech platform.