Projects
A selected portfolio across AI research, software systems, speech and DSP, FPGA design, and earlier hardware work.
UROP-funded research
An ongoing research prototype connecting speech recognition, transcript processing, document generation, and validation mechanisms for safer AI-assisted clinical documentation.
Deep learning + DSP
Led a four-person team building an encoder-decoder that combines residual convolutions, temporal modeling, frequency attention, and a learned noise gate.
Sanitized experience case study
Improved a production automation platform across its TimescaleDB queries, React interface, caching layer, and deployment workflow.
Digital hardware
Modular RTL that accepts standard 8-bit ASCII and emits variable-length Morse sequences while keeping the design in one 100 MHz clock domain.
Local document search
A tested local search platform with document ingestion, Qwen embeddings, Qdrant retrieval, and CLI, API, and browser interfaces.
Computer vision
A pose and movement-assessment prototype that maps three-dimensional landmarks to exercise-specific posture signals without making a clinical diagnostic claim.
Earlier engineering work
An ergonomic split keyboard project spanning KiCad board and plate design, firmware adaptation, team leadership, and competition presentation.
Deployed systems
A smaller Telegram project combining live weather, semantic wardrobe retrieval, Docker deployment, and Prometheus/Grafana observability.