Vahit Uzunlar
I’m an AI Engineer and AI Systems Architect based in Leiden, taking AI systems from business requirements all the way to evaluated, production-grade deployments.
About
I’m an AI Engineer and AI Systems Architect based in Leiden, taking AI systems from business requirements all the way to evaluated, production-grade deployments. I’ve spent the last five years building applied LLM solutions end-to-end—ranging from self-hosted GPU inference on tensor-parallel clusters and multi-tenant RAG, to multi-agent orchestration with LangGraph and MCP—then wrapping everything in production product layers with Django/FastAPI and modern web frontends.
I enjoy owning both the technical and commercial sides of AI: designing evaluation workflows, selecting models through empirical testing and fine-tuning, and building GPU cost models and optimization plans so systems are feasible at scale. I’ve led enterprise AI platform work for large subscription businesses, built multi-agent products for legal and clinical use cases, and delivered student-teacher assistant platforms—often as the sole architect or team lead, with direct client collaboration to turn requirements into reliable releases.
- 13 years' experience
- Joined October 2026
Skills· 16
Languages
- EnglishAdvanced
- TurkishFluent
- DutchIntermediate
Industry experience
- Software & Internet
- Government
- Healthcare
- Professional Services
- Education
- Financial Services
Work experience· 8 roles
Delivered a production student assistant serving 700 users across students and teachers, from business requirement to full release. Built multi-tenant RAG on Elasticsearch with isolated per-tenant index access. Led ML model training and deployment with BentoML and guided a team of five engineers through delivery. Used FastAPI, Django, Next.js, MongoDB, PostgreSQL, Redis, Azure, and Docker.
Built a multi-clinic dietitian agent with 12 custom tools to support day-to-day clinical dietitian tasks. Implemented a constraint-based optimization routine for diet plan generation. Developed a multi-tenant backend (Django + FastAPI services) using MongoDB/PostgreSQL/Redis and Stripe billing, delivered within a Docker/Nginx deployment environment, and led client meetings to translate requirements into production-ready releases.
Developed and scaled an AI-native multi-agent legal research platform with grounded access to online legal sources, including national legal codes and regulations, Turkish precedent, Council of State decisions, EU legislation, German federal law, and US federal/state law. Implemented the stack with LangGraph, Next.js, Redis, PostgreSQL, Stripe, OAuth, Docker, and Nginx.
Education
Hire Vahit Uzunlar today
Post a job and invite Vahit, or contact them directly. No hidden commission.