I've built RetailMind, an event-driven retail analytics platform that combines YOLOv8, ByteTrack, Django, React, Redis, and Gemini API to turn video occupancy events into business summaries.
I designed its benchmarking and observability framework to measure end-to-end latency, backend requests, event processing, serialization, WebSockets, CPU, and RAM, establishing a ~36 ms mean event-latency baseline in simulation. I also containerized the full stack and validated the detection-to-summary pipeline on recorded footage.
I've also built a 98% accurate Braille recognition and audio-conversion pipeline, and independently delivered a production B2B website for Ross Enterprises from requirements through deployment and post-launch support.
