At Sunstone, I engineered a Python ETL pipeline that cut data-processing latency by 94%, replacing manual CRM sync with an automated, self-serve data flow. I also maintained 99.9% data integrity across enrollment funnels with automation and feedback-loop validation.
At Sunstone, I deployed an LLM-powered automation layer for ticket triage, reducing manual triage time by 45% through intent detection and automated routing.
I built MyHealthApp, an AI-powered healthcare assistant with a RAG pipeline serving 200+ users at sub-800ms p95 latency. I added an evaluation harness and hallucination guardrails, cutting unconstrained generation paths by 45%.
On MeaningIt, an on-device AI translation engine, I built OCR and speech-to-text inference for 500+ users, cutting response time by 60%. I also implemented caching and deduplication, reducing redundant inference calls by 30%.
