What I've Built
Projects
Building at the intersection of AI, finance, medicine, and sports.
January 2025 – Present
PoliStock
Financial Intelligence Platform
FinTechSaaSAIReal-time SystemsNLP
- Financial intelligence SaaS monitoring 20+ volatility sources 24/7, analyzing 5,000+ stocks with real-time alerts in under 1 minute, giving retail traders institutional speed.
- Pioneered Natural Language Orders (NLO): users set buy/sell triggers in plain English, auto-executed across 27+ brokerages including Robinhood, Fidelity, and Charles Schwab.
- Backed by Gener8tor's gBETA; scaled to 200+ active users, $250K+ invested via platform insights, 4.8-star rating across 200 reviews at $9.99/month.
- Integrated SEC filings, Reddit, CNBC, NY Times, and AI-powered event monitoring with bank-level encryption for brokerage access.
August 2022 – Present
Spectator AI
Sports Analytics Platform
Computer VisionDeep LearningAzureMLOps
- Secured $25K in Azure credits and $2.5K in OpenAI API credits; architected scalable ML infrastructure on Azure ML with Docker containerization.
- Implemented a computer vision pipeline with YOLOv11 for player tracking achieving 95% mAP; built transformer-based models with BERT fine-tuning for sports content analysis.
November 2024 – Present
Trauma THOMPSON
AI Copilot for First Response Medicine
Computer VisionMixed RealityPyTorchHoloLensActive Learning
- Built YOLOv11 CV models for medical tools with 96% mAP@[0.5:0.95]; quantized models and focal loss to handle critical medical data imbalances.
- Slashed data annotation needs by 80% with an active learning pipeline using uncertainty sampling, distributed with PyTorch + Ray.
- Crafted a mixed reality system for HoloLens to visualize ML inferences in real time; contributed statistical analysis to peer-reviewed research papers.
November 2024 – Present
NASA Glacier Tracking System
Climate Science & Geospatial ML
Climate ScienceGeospatialPythonCUDABig Data
- Developed a glacier tracking system at NASA Goddard Space Flight Center using satellite imagery and geospatial ML models.
- Leveraged CUDA-accelerated processing pipelines to analyze large-scale climate datasets, enabling real-time glacier extent monitoring.
- Contributed to climate change research with reproducible big-data workflows that scale across decades of satellite archives.
2023
AMP Parkinson's Disease Progression
Biomedical ML Research
Random ForestsNeural NetworksStatistical AnalysisResearch
- Built ensemble models (Random Forests + Neural Networks) to predict Parkinson's disease progression from protein and peptide biomarker data.
- Applied rigorous statistical analysis: feature selection, cross-validation, and SMAPE evaluation, achieving competitive Kaggle leaderboard scores.
- Explored interpretability techniques to surface clinically meaningful biomarkers relevant to disease staging.
2023
IPL Cricket Index
Sports Data Analytics
Data AnalyticsExcelData ScienceData Visualisation
- Constructed a comprehensive IPL performance index aggregating batting, bowling, and fielding metrics across multiple seasons.
- Built dynamic Excel dashboards with pivot tables and custom visualizations to surface player and team trends for non-technical stakeholders.
- Identified under-valued players through statistical scoring models, providing actionable insights for auction strategy analysis.
2022
FIFA Predictions
Soccer Match Outcome Modeling
Data AnalyticsExcelData ScienceData Visualisation
- Modeled FIFA World Cup match outcomes using historical team performance data, ELO ratings, and tournament context features.
- Built data pipelines in Excel and Python to clean, transform, and visualize international football statistics across multiple tournaments.
- Achieved strong prediction accuracy on group-stage fixtures, outperforming baseline Elo-only models through engineered features.
2022
Covid-19 Spread Analysis
Epidemiological Data Science
Data VisualisationClusteringCorrelationsBayesian Analysis
- Analyzed global Covid-19 spread patterns using clustering (K-Means, hierarchical) to group countries by transmission trajectory.
- Applied Bayesian analysis and correlation studies to identify socioeconomic factors most predictive of case growth rates.
- Created interactive visualizations mapping outbreak timelines, enabling intuitive comparison across regions and policy regimes.
2022
Zillow Housing Prices
Real Estate Predictive Modeling
Linear RegressionStatistical AnalysisNeural Networks
- Built linear regression and neural network models to predict Zillow housing prices from location, square footage, and market features.
- Performed statistical analysis: outlier detection, multicollinearity checks, and residual diagnostics to validate model assumptions.
- Compared OLS, Ridge, and MLP models across train/test splits, with neural networks outperforming linear baselines by 12% on RMSE.