Istanbul, Türkiye
Computer Engineering student at METU (graduating June 2026) with hands-on experience in quantitative finance, low-latency systems, and distributed data engineering. Built Rust/Python pipelines for real-time market data ingestion at a US-based quant trading firm; self-initiated tooling that identified a critical production bug causing live P&L losses. Conducted cross-chain time-series research (Granger causality, cointegration) and contributed to crypto market-making execution infrastructure. Architected and benchmarked distributed ETL pipelines (Kafka, Airflow, Dask) across multiple parallelization strategies, cutting end-to-end processing time by 43%. Conducted HPC scalability analysis on LRZ SLURM clusters, achieving 7.6x write and 5.1x read speedups with 64 workers.
• Authored an ACM DEBS 2026 paper on a dataset-specific ML framework (association rules, decision trees, linear regression) for Dask configuration tuning — achieving R² = 0.94 and 91.3%-confidence rules over 494 benchmark configurations. • Architected and benchmarked an end-to-end distributed ETL pipeline (CSV → Kafka → PostgreSQL → Kafka → Visualization) using Apache Airflow and Dask; compared five parallelization strategies, cutting end-to-end time by 43% (40:30 → 23:19). • Benchmarked Dask distributed (1–64 workers) on the LRZ 64-CPU SLURM cluster with PostgreSQL as the storage backend, achieving 7.6x write and 5.1x read speedups by tuning partition-to-worker ratios and resolving WAL fsync contention.
• Self-initiated a Python backtest comparison tool after observing backtest-to-live divergence; integrated with AWS RDS to analyze trade timing, PnL, and position sizing — identified a critical production bug and stopped ongoing losses. • Built low-latency Rust pipelines ingesting WebSocket tick feeds, enriching them with volatility and volume-bucketing features; published via Redis Pub/Sub for real-time consumption by trading algorithms. • Eliminated Redis connection churn via pooling, removing per-message reconnect overhead and reducing end-to-end latency. • Corrected volatility resampling and column mappings, fixing upside bias; post-fix volatility feature importance rose 50% in decision-tree models. • Conducted time-series correlation analysis (Granger causality, cointegration) on Solana and Avalanche data; identified a 3-day AVAX→SOL predictive relationship and 1-week Google Trends leading indicators (R² = 17.3% for SOL).
• Developed async Python trading bots utilizing asyncio and aiohttp for market-making strategies. • Diagnosed and resolved order-cancel lag spikes, enhancing the efficiency of live operations. • Engineered real-time crypto ETL pipelines, integrating tools like Prometheus and Grafana for monitoring.
• Developed an AI chatbot using OpenAI APIs to assist users with form creation and editing. • Integrated the chatbot into a Streamlit frontend for real-time conversational interaction. • Employed iterative prompt engineering and few-shot learning to minimize hallucinations and off-topic responses.
• Delivered web scraping and automation projects for clients in e-commerce, social media, and government sectors. • Utilized Scrapy, Playwright, and aiohttp to build robust scraping solutions. • Reverse-engineered undocumented Twitter/X API endpoints for social media analytics. • Developed async scrapers for major platforms, ensuring compliance with rate limits and anti-bot measures.