At University of Georgia, I developed specialist models for smart-home activity recognition across five CASAS testbeds, improving accuracy while reducing compute costs to one-sixth of the strongest competing system. This work anchored an IEEE PerCom 2027 paper.
At Stylabs AI Native Studio, I shipped production features across four live web apps and rebuilt complex MongoDB aggregation pipelines, cutting data-heavy screen wait times by 50%. I also established the technical foundation for the Profound Properties platform and built an AI-assisted data-seeding pipeline.
At Lykan Technologies, I optimized customer-facing endpoints with Redis caching and N+1 query fixes, reducing average latency by 61%. I also built a GitHub Actions and Docker pipeline that shortened release cycles while maintaining zero-downtime deployments. On my HomeSense project, I turned my thesis model into a deployable product with live Kafka streaming, anomaly detection, and a floorplan digital-twin viewer.
