AI Engineer × Quant · ML Benchmark Engineer at Datacurve (YC) · ex-derivatives trader · Kaggle Expert (top 1%)
🌐 Portfolio · 📈 Quant view · 📧 Email · 💼 LinkedIn · 📊 Kaggle
I build AI systems that are measured, not guessed. That means LLM apps, agents and the benchmarks that evaluate them, plus quant research that holds up out of sample.
- 🧪 ML Benchmark Engineer @ Datacurve (Y Combinator): I design Kaggle-style challenges (data pipeline, grader, rubrics, reference solution) used to train and evaluate autonomous AI agents, and calibrate them so frontier agents score below 0.8
- 📈 2.5 years trading NIFTY / BANKNIFTY weekly options on a live desk; 12th in Germany in IMC Prosperity 4 (230 / 18,803)
- 🛠️ Ships full-stack: TypeScript / Next.js / FastAPI / PostgreSQL (QuantPilot: AI copilot + quant-ranked copy trading on Hyperliquid)
- 🎓 B.Sc. International Business Information Systems, Furtwangen University (grade 1.55)
- 🎯 Open to AI/ML engineering & quant internships / Werkstudent roles in Germany
AI / LLM: Claude & OpenAI APIs · RAG · Embeddings · Vector DBs · Agents / Tool-use · LLM evaluation · Benchmark design · Fine-tuning ML / Data: Python · PyTorch/scikit-learn · Pandas · NumPy · Jupyter · Spark Software / Deploy: TypeScript · Next.js · FastAPI · PostgreSQL · Docker · Streamlit · Vercel Quant Finance: Derivative pricing · Stochastic calculus · Options & Greeks · Black-Scholes / Heston · GARCH · VaR / CVaR · Monte Carlo · Fixed income · Advanced Excel / VBA
- 🥈 Kaggle Competition Silver Medalist — top 4% (120 / 3,325 teams), CIBMTR survival prediction
- 🏆 WorldQuant BRAIN — Gold Level · 🎓 Applied Data Science Lab, WorldQuant University
- 📜 NISM (SEBI) — Equity Derivatives & Research Analyst certifications
- 💹 AlgoTrade Hackathon 2025 (Zagreb) · MIT OpenCourseWare Financial Mathematics
| Project | What it does | Stack |
|---|---|---|
| 🤖 Financial Filings RAG | Chat with SEC 10-K/10-Q filings using retrieval-augmented generation | Claude · RAG · Vector DB |
| 🧠 Financial Analyst Agent | AI agent that analyzes stocks by autonomously calling market-data & risk tools | Claude · Agents · Tool use |
| 🚦 Traffic Sign Recognition | PyTorch CNN classifying German traffic signs (GTSRB, 43 classes) at 97.7% accuracy | PyTorch · Computer Vision · CNN |
| 📊 Portfolio VaR Analyzer | Value-at-Risk analysis for investment portfolios | Python · Finance |
| 📈 Volatility Curve Prediction | Data-imputation competition: predicting missing volatility values | ML · Jupyter |
| 🧮 Black-Scholes vs Heston | Comparing option-pricing models on real data | Python · Quant |
| 📗 Excel VBA Risk Toolkit | Macro-driven portfolio risk report (VaR, Sharpe) with OOP VBA & charts | Excel · VBA · Finance |
I'm always happy to talk about AI engineering, applied ML, and quantitative finance.