About
My name is Eugene Zhulikov. I build ML services and end-to-end document recognition pipelines:
LLMs, YOLO, OCR, field extraction, NLP, benchmarking and model comparison.
I focus on production-grade delivery: containerization, CI/CD, protocols, observability, and deployments
on-prem and in cloud environments.
Recent experience (focus)
- Designed and developed end-to-end document recognition pipelines (LLM + YOLO + OCR).
- Fine-tuned Llama and trained YOLO; ran experiments to improve extraction quality.
- Benchmarked models and performed comparative analysis to pick optimal solutions.
- Built scalable services (Docker, FastAPI, Protobuf, CI/CD).
- Deployed and maintained models in production (on-prem & cloud).
Earlier
- Gaming/odds computation, analytics services, ETL tooling, and process automation.
- P2P crypto exchange automation: bots/integrations improving turnover and UX.
- Freelance projects: Telegram/Discord tools, automation, small product work.