Consulting
Consulting
Technical leadership and hands-on delivery for growth-stage and enterprise teams. Selected engagements.
Pharmaceutical
AbbVie: An AI agent for safety label review
AbbVie’s medical safety team reviews product labels of 10,000 to 40,000 words and codes every adverse event with MedDRA terminology, tracking the work in spreadsheets. It consumed physician time and carried compliance risk.
I helped build an LLM-based agent with a custom semantic re-ranker that extracts adverse events and suggests MedDRA codes, with human experts verifying the output before it reaches the safety system.
- 150–600 physician hours saved per year
- 88% average recall on 10 labels
- Nearly 100% MedDRA exact match
- Found events missed in manual review
Patent intelligence
Patent Forecast: Removing the ceiling on subscription growth
Patent Forecast turns patent data into interactive analytics, but legacy ML models that relied on supervised learning and low-quality data capped the business at about 15 new subscriptions a month.
I fixed the immediate gaps, evaluated what worked, and then redesigned the data pipeline from the ground up, which removed the dependency on supervised learning.
- $600K additional ARR
- Hundreds of subscriptions a month, up from about 15
- Subscription creation 40x faster
- Same headcount
Note
Looking for the same kind of help?
I’m not taking on new client engagements, but I’m glad to talk about advisory work.