Resume.

• Leading migration of the FinGas analytics platform (100+ traders across NY, London, and Düsseldorf) to Azure Container Apps with GitHub Actions CI/CD; powers live positions, P&L, and fundamentals dashboards driving gas trading decisions.
• Built Python/SQL pipelines in Snowflake ingesting millions of rows daily from vendor feeds (Enverus, LSEG, Kpler) covering production, storage, and pipeline flows, underpinning the desk's view on basis spreads and seasonal positioning.
• Developed ML-based anomaly detection across 216 North American gas pipelines, flagging flow shifts the desk investigates for basis and locational trading; built a backtesting framework to validate seasonal strategies before deploying capital.
• Built AI research agents (LangGraph, MCP) compiling daily intelligence on key oil-producing regions, surfacing supply dynamics that influence global LNG flows and US gas exports.

• Built Python/Flask microservices integrating Azure, Snowflake, and Excel to deliver curve construction, spread analytics, and scenario modeling tools used by gas analysts for daily pre-market prep and trade ideation.
• Re-architected the platform's data layer with Redis caching and async task queues backed by Celery workers, cutting peak-hour page loads from ~60s to under 2s and unblocking real-time use during the morning open.

• Conducted research on market microstructure, focused on limit order book dynamics and execution behavior.
• Built C++ components for the lab's limit order book simulator, running experiments on order arrival and matching.
• Analyzed historical tick data in Python to characterize spread, depth, and volatility across equity and futures.

• Built backend services on the subscription and checkout teams, two revenue-critical paths in Getir's platform (the quick-commerce decacorn later acquired by Uber). Cut subscription renewal latency from ~2s to ~100ms by offloading read-heavy aggregations to AWS Redshift.
• Engineered a server-side alerting system using a task manager pattern and Slack notifications, dropping unidentified error rate from 22% to 3.4%. Refactored Chain of Responsibility in checkout to speed payment and promotion rule rollout.

• GPA 3.7 / 4.0. Dean's List (Grainger College of Engineering), 5 semesters.
• Phi Kappa Theta Gregory Wooters Academic Excellence Scholarship: AY 2021–22, 2022–23, 2023–24.
• Bloomberg Certifications: Market Concepts (BMC), ESG, Spreadsheet Analysis.

• Led a 4-person team building a real-time ADS-B tracker ingesting live aircraft telemetry into a Cesium 3D globe, rendering thousands of concurrent flight paths with sub-second update latency.
• Implemented Closest Point of Approach (CPA) collision prediction scoring horizontal/vertical separation and time-to-conflict, surfacing live conflict alerts ranked by severity across the active airspace.
• Designed heuristic and ML-based anomaly detection for GPS spoofing and implausible flight behavior, flagging unrealistic vertical rates, heading changes, and position jumps in real time.

• Built a PyTorch multi-task framework on MTI-Net jointly learning depth estimation, semantic segmentation, and edge detection on NYU Depth V2, with custom guided attention routing segmentation and edge features into the depth head.
• Designed a structured depth loss combining L1, gradient, and edge-aware regularization terms, achieving 0.0323 MAE / 0.0416 RMSE on depth and 62.17% pixel accuracy on segmentation.

A fake-news classifier on the LIAR dataset built with classical IR methods rather than a transformer. The pipeline stitches TF-IDF retrieval, LDA topic modeling blended through a Gaussian Mixture, and RoBERTa-derived sentiment features into a logistic-regression head you can actually inspect.

• Drove chapter GPA up 15% over two years (highest in chapter history).
• Co-administered a $100,000+ annual operational budget across membership, events, and facilities.