I'm a full-stack engineer, frontend-leaning by background, moving deliberately into AI engineering. I build products end-to-end in TypeScript — database, API, UI — and these days most of that time goes into the AI layer: agent loops, tool-calling, retrieval, and evaluation. I like to understand these systems from first principles rather than wiring up a library, and I approach the work with a product mindset — putting on that hat and building things that solve real problems.

• 🧩 Frontend Tech Lead @ Sanstech — building and maintaining large-scale web applications, and driving the company's internal AI transformation by weaving AI into everyday workflows.
• 🧪 Going deeper on AI, in the open — RAG and GraphRAG, information extraction, fine-tuning with the HuggingFace/PyTorch stack, and evaluation as a first-class step. Working notes live in the garden.

• 🔎 Code RAG — a retrieval system for codebases: query-shape routing between BM25 and dense retrieval, AST-aware chunking with tree-sitter, BGE-M3 multilingual embeddings for Turkish questions over English code, Milvus for storage, and confidence bands that let it abstain instead of guessing. Every decision measured, not assumed.
• 🧠 porti-code — an open-source coding agent built from scratch, Claude Code-style. I built it to understand agentic systems from first principles: the agent loop, tool-calling, and LLM orchestration — not just consuming an SDK.
• 🤖 AI in production — multi-provider LLM pipelines, MCP tool-use, and retrieval with embeddings and pgvector, shipped to real users.
• 🏗️ End-to-end internal product — designed and built it alone across frontend, backend, and DevOps, with a particular focus on the AI layer.
• ⚙️ Large-scale migrations — led a JavaScript → TypeScript migration and an scss/styled-components → Tailwind transition, improving maintainability, performance, and scalability.
• 🎨 Full-stack in TypeScript — semantic HTML and accessibility on the frontend, clean API and data design on the backend.

mindmap
  root((fport))
    Frontend
      React · Next.js · Vue
      Tailwind · shadcn/ui · Radix
      TanStack Query · Zustand
      Semantic HTML · a11y
    Backend
      Bun · Node · Hono
      Drizzle · Better Auth · Zod
      BullMQ · Redis
      Postgres · MySQL · MongoDB
    AI
      Vercel AI SDK · MCP
      Anthropic · OpenAI
      RAG · embeddings · pgvector · Milvus
      HuggingFace · PyTorch · LoRA
      Evals · tracing · Langfuse
    Infra
      Cloudflare Workers · D1 · R2
      Docker · Nginx · AWS · Vercel
      Vitest · CI/CD

The tools I actually keep an eye on live here → github.com/stars/fport/lists/stack

• ✨ porti-code — open-source coding agent
• ✨ Jrontend — micro-frontend tool
• ✨ React SSR Boilerplate

Working notes, written while building — not after.

• 🌿 AI Garden — RAG vs. fine-tuning, chunking, hybrid search, agentic retrieval patterns, LoRA/QLoRA, and how to measure any of it
• 🔎 Code RAG — building retrieval for a codebase: what worked, what measurably didn't
• 🎤 Experience & Talks — talks, ventures, and the courses I've written
• 📘 Cypress Notes (Turkish)
• 📝 Undefined and Null
• 🐍 YazYap — Python tutorial for high school students

• LinkedIn — https://www.linkedin.com/in/furkanportakal
• Twitter/X — https://twitter.com/getporti
• Email — furkanportakalx@gmail.com
• Personal Website — https://www.getporti.com