Shipping production ML & LLM systems for large-scale humanitarian operations at Türk Kızılay (Turkish Red Crescent). I own impact measurement for the world's largest cash assistance program (1.5M+ beneficiaries) and build AI that turns raw relief data into verifiable, source-cited decision-making tools.

• LLM & Agentic AI — LangGraph multi-agent pipelines, RAG, fine-tuning (LLaMA/GPT), LLM validation
• Machine Learning — Transformers, FT-Transformer, self-supervised learning, XGBoost/LightGBM/CatBoost, SHAP/LIME
• Causal Inference & Statistics — A/B testing, panel data, econometrics, Type I/II error analysis
• MLOps — Docker, GCP, MLflow, Weights & Biases, CI/CD, Streamlit

Live multi-agent LLM intelligence platform serving humanitarian organizations — automated SITREPs, donor proposal wizard, and weekly bulletins across 30+ countries, with 8 integrated data sources and source-verified outputs.

🔗 sightlinehumanitarian.com

• sightline — Agentic system for SITREPs, secondary reviews, weekly bulletins, proposal writing
• proposal — Sightline Proposal Design Studio (ECHO/HIP, USAID/BHA, OCHA CBPF)
• Targeting_w_Transformer — Self-supervised FT-Transformer for vulnerability prediction
• More experimental work in mynotes, mypersonalagent, and various projects

Python · PyTorch · FastAPI · Flask · SQL · LangGraph · ChromaDB · RAG · Hugging Face
Docker · GCP · MLflow · W&B · Git · CI/CD · Streamlit · SQLite · GeoPandas · QGIS · ODK

• 💬 LinkedIn — open to Senior Data Scientist / AI Engineer / ML Engineer roles (remote-friendly)
• ✉️ Email — [email protected]
• 🌐 Website — serkankizilirmak.com