AI Engineer
Burak Emre Tüylüoğlu
I build AI systems that run in production — retrieval at scale, agentic pipelines, and the unglamorous work of keeping them reliable.
Open to new roles — most recently AI Engineer at CodLab AI, through Jul 2026.
MSc Computer Engineering, Karabük University — thesis: large-scale academic author disambiguation
What I build
Entity resolution at scale
Sole engineer on an academic author disambiguation pipeline over bibliometric data, designed and run in production, architected toward 85M+ articles (roughly 500M author mentions) on a single machine.
~92% accuracy · 200-case ground-truth benchmark
Making pipelines fast
Cut full-pipeline runtime on a 2.5M-record dataset roughly in half; the largest win was replacing an O(N²) read pattern with O(N) keyset pagination.
~4h → ~2h
Retrieval & agents
Embeddings, vector search, RAG and GraphRAG; agentic pipelines with LangChain and LangGraph, tool calling, MCP.
Demos
Everything here runs in your browser — no API calls, nothing leaves the page.
01
client-side · no API · works offline
Disambiguation playground
Two author records, nine decision levels and a merge cap. Watch which signals fire, which one vetoes, and why the final call was made.
Open →
02
client-side · BM25 vs semantic proxy
Keyword vs semantic retrieval
One query, two rankings over passages about my own projects — where keyword search wins, where it misses, and how a cited answer gets assembled.
Open →
03
simulated trace · step-by-step
Agentic trace viewer
Replay an agent loop step by step: which tool it picks and why, retrieval hits, the step budget counting down, and where the time and tokens go.
Open →
Recently
- CodLab AI — Artificial Intelligence EngineerFeb 2025 – Jul 2026
- Mitsubishi Electric Innovation Center — AI InternSep – Nov 2024
- DIATICS — AI InternMay – Sep 2024