A deep dive into my architectural choices, machine learning models, and system designs.
01 / SaaS Platform
SQLBuddy
A text-to-SQL SaaS platform enabling non-technical users to query company databases in natural language. Built with a multi-tenant architecture and strict role-based access control (RBAC).
Designed and built end-to-end a hybrid rule-based and ML anomaly detection system for a multi-channel IoT pipeline processing ~60,000 sensor records (pH, ORP, conductivity). Benchmarked LightGBM and XGBoost models.
FastAPIMachine LearningPostgreSQL
PROPRIETARY PROJECT (LIVE) 🔒
03 / Neuroevolution Simulation
TrackLearner AI
A self-driving car simulation where 60 agents evolve over up to 150 generations using NEAT neuroevolution. Features 5-beam radar perception, OpenCV-based track processing, and real-time network visualization at 60 FPS.