Prop Trading Performance Dashboard — A desktop analytics platform I built to track my own trading performance across multiple prop firm accounts. Ingests raw trade CSVs, aggregates fills using interval overlap analysis, computes per-day statistics (expectancy, profit factor, drawdown), and models firm-specific payout eligibility rules in real time. Built end-to-end with TypeScript, React, SQLite, and Electron. Currently building two major features:
- Analytics Engine — Algorithmic analysis over the user's trade history to surface performance patterns: optimal time-of-day windows, position sizing vs. P&L, trade duration profiling, win rate by day of week, and behavioral tendencies after streaks and drawdowns. All computed locally through statistical algorithms — no AI dependency.
- AI Assistant — An LLM-powered layer (GPT-4) with full context over the user's trading data. Includes an automated performance review that identifies strengths, weaknesses, and actionable improvements based on the analytics output, plus a conversational chat interface for querying trade history in natural language (e.g., "How do I perform after a red day?" or "What's my expectancy on trades under 2 minutes?").
| # | Project | Description | Tech |
|---|---|---|---|
| 1 | Prop Trading Dashboard | Desktop analytics platform for tracking prop firm trading performance — trade aggregation, per-day statistics, payout eligibility modeling, and interactive P&L curves | TypeScript, React, Electron, SQLite, Recharts |
| 2 | Rotten Tomatoes Predictor | Predicts Rotten Tomatoes scores from raw screenplay text using BERT fine-tuning and a fusion architecture combining transformer embeddings with numeric features | Python, BERT, PyTorch, NLP |
| Repository | Description |
|---|---|
| Kaggle Challenges | Kaggle competition entries — NLP Disaster Tweets, Monet Style Transfer with GANs (Rank 20), Histopathologic Cancer Detection |
| CU Boulder MSDS | Coursework from my MS in Data Science at the University of Colorado Boulder |
Languages: Python, SQL, R, TypeScript ML & Statistics: scikit-learn, TensorFlow, Pandas, NumPy, hypothesis testing, regression, ANOVA, experimental design Data Engineering: Relational databases, ETL pipelines, query optimization, data validation Visualization: Power BI, Tableau, Recharts, dashboard development Tools: Git, Jupyter, Vite, Electron
Before data science, I spent two decades in high-stakes operational roles — leading diplomatic security missions at the U.S. Embassy in Baghdad, managing infantry squads through combat deployments, and coordinating cross-functional teams where failure wasn't an option. That background shaped how I approach data problems: structured planning, rigorous validation, and clear communication of results.