At Lundbeck, I built a production multi-call LLM pipeline that generated HCP-tailored drug brochures from approved product content, cutting sales-rep prep time by 40% across 460 reps. I also engineered data pipelines processing more than 10 million records daily, reducing data latency from three hours to 15 minutes.
At Airriva, I developed pricing and demand models that improved profits and pricing accuracy, and mapped occupancy patterns across 4,500+ listings. More recently, I built AI projects including a guarded natural-language-to-SQL assistant and a multi-agent trading copilot; at Omdena, I led ML work on fraud detection and rooftop solar-panel detection.
