Projects
Squary AI
Squary AI is legal AI that learns how a law firm produces its work and delivers that expertise consistently on every new matter. I co-founded it in January 2025, and it grew out of SQUARE, an enterprise question answering system I architected at Cognistx.- Firm-approved work product: summaries, chronologies, and demand letters built around each firm’s own workflows, with source-linked citations for every claim
- Millions of pages processed for mid-market firms in personal injury, workers’ compensation, employment, and immigration, with document review up to 70% faster
- Custom LLM orchestration that splits large jobs into tracked steps, producing structured outputs such as 5,000+ row compliance matrices
- Hybrid retrieval (keyword + vector search) on MongoDB Atlas, at 10x the data volume and lower cost (case study)
- Security-first, multi-tenant architecture with VPC isolation, encryption, two-factor authentication, and full audit trails
- Integrations with SharePoint, Filevine, Google Drive, OneDrive, Dropbox, and Amazon S3
MoleculeAI
As Lead Data Scientist for MoleculeAI, an AI drug discovery platform at Cognistx, I led development of graph neural network models to support the discovery of medicinal psilocybin derivatives.- Built a molecular property prediction repository that automated model tuning and evaluation, which sped up the discovery pipeline
Product Placement
As Data Scientist on Product Placement, a proof-of-concept project at Cognistx, I built a tool that used Stable Diffusion to generate realistic backgrounds, shadows, and reflections for retail product imagery.
- Designed a serverless AWS architecture for low-cost inference APIs, demonstrating feasibility at scale