Dogu Cagri O.

Full Stack AI Developer | React, Next.js, Node.js | LLM, RAG, Agents

AI Product Engineer & Full Stack Developer (React, Next.js, Node.js, TypeScript). I ship LLM features, RAG and AI agents inside production SaaS — with the reliability layer (guardrails, retries, evals) that keeps AI from breaking in front of users. AI Product Engineer & Full Stack Developer (React, Next.js, Node.js, TypeScript). I ship LLM features, RAG and AI agents inside production SaaS — with the reliability layer (guardrails, retries, evals) that keeps AI from breaking in front of users. Most teams don't struggle to call a model API. They struggle with everything around it: hallucinated answers, prompt changes that silently regress, token costs that creep, rate limits, and demos that never survive real users. That's the part I own. 🔧 WHAT I BUILD ● LLM features inside existing products — summaries, extraction, analysis, AI-assisted workflows (OpenAI + Anthropic Claude APIs, structured outputs, tool calling) ● RAG done properly — PostgreSQL/pgvector, chunking, hybrid search, answers with citations instead of hallucinations ● AI agents with guardrails — approval gates, scheduled background execution, monitoring (I run my own agents in production) ● The plumbing — REST APIs, webhooks, queues, idempotent retries, graceful failure handling ● The interface — React, Next.js, Vite, TypeScript, Tailwind CSS, React Native (Expo) 🏆 TRACK RECORD ● Elsevier (Osmosis): enterprise Next.js migration, test coverage held above 90% ● HeliosX Group (ZipHealth, MedExpress, Dermatica): 2 years of patient-facing product at Europe's largest telehealth group, every change shipped behind Amplitude A/B experiments ● Parrot AI: joined a high-traffic consumer AI platform and shipped production frontend within days ● My own AI products — Sidely (hosted background agents), Audiancer (AI audience intelligence), Qwispr (real-time AI call assistant: Claude + Deepgram streaming) — so the patterns I ship for clients are ones I operate myself ● Upwork: 5,400+ hours, 100% Job Success, Top Rated Plus, including a 1,162-hour engagement with a leading Scandinavian CRO agency 🧰 STACK React · Next.js · Node.js · TypeScript · JavaScript · Vite · Tailwind CSS · React Native (Expo) · PostgreSQL · Supabase · Firebase · Redis · Docker · AWS · Vercel · Stripe · OpenAI · Anthropic Claude · pgvector · MCP I use Claude Code and Cursor daily — and I review AI-generated code the way I review any code: with tests and a diff, before it ships. ✅ Great fit: you have a live product (or a funded build) and AI features you want shipped and owned end to end — reliable, not just demoed. ❌ Not a fit: a throwaway chatbot demo. 🕐 30+ hrs/week · full EU overlap · 4+ hours daily overlap with US East Coast (Izmir, UTC+3). Tell me which manual step in your product should disappear, or which AI feature is stuck at demo stage — I'll tell you honestly whether AI is the right tool for it, and exactly how I'd ship it.