# Paul Mangiamele, PhD

> Making AI trustworthy for the people who need it most.

AI Product Design Leader with 18+ years shipping production AI and healthcare products. Computer scientist, HCI researcher, bench scientist, computational biologist, designer, and engineer. San Francisco, CA.

## Current Role

**Member of Technical Staff** — AWS Applied AI (2025–Present)
Building AI systems for drug discovery scientists across academia and biopharma — protein design, antibody engineering, and clinical innovation. 90+ user interviews across Harvard Medical School, Stanford, Broad Institute, Yale, Columbia, and biopharma partners. 60%+ production code authored (Python, React, TypeScript). Research focus: multi-agent human-AI collaboration in drug discovery.

## Previous Roles

- **Product and Design Lead, Automotive** — AWS (2022–2025): Co-developed autonomous vehicle platform with Continental/AUMOVIO partnership. Petabyte-scale sensor data processing, AI agents, 60+ stakeholder interviews.
- **Principal UX Designer** — Roche/Genentech (2015–2022): First product/UX hire. Built NAVIFY clinical genomics platform (10,000+ physicians, 500+ clinical labs). CE-IVD certified Mutation Profiler. 75% reduction in variant curation time.
- **Founder** — UX Bio / GeneLinea (2014–2016): Genomics platform for biologists to run computational pipelines without coding.
- **Head of Bioinformatics** — Iowa State University (2012–2014): Led computational biology research team.

## Education

- PhD, Human-Computer Interaction — Iowa State University (2008–2014)
- BS, Computer Science & Philosophy — Iowa State University (2003–2007)
- 10+ years wet lab experience (DNA isolation, PCR, sequencing, library prep)
- 12+ peer-reviewed publications

## Key Projects

### PocketScout: Binding Site Intelligence (2026)
Open-source MCP server for evaluating druggable binding pockets on protein targets. Six domain-informed tools composing across five public APIs (UniProt, PDB, AlphaFold DB, ChEMBL, PubMed). Tool descriptions engineered as prompts to guide Claude's scientific reasoning.
- GitHub: https://github.com/Proprius-Labs/pocketscout-mcp
- Page: https://www.paulmangiamele.com/projects/pocketscout

### Calvin: Intelligent Tab Management (2024–Present)
AI-native Chrome extension that captures and organizes everything you encounter online. Built end-to-end with Claude Code. Multi-platform: Chrome extension, web dashboard, iOS app.
- Website: https://calvintabs.com
- Chrome Web Store: https://chromewebstore.google.com/detail/ebalblfbmadhcpeckopbckgpggbbheck
- Page: https://www.paulmangiamele.com/projects/calvin

### TaxonSphere: Phylogenetic Analysis Platform (2024)
GPU-accelerated phylogenetic visualization platform rendering 1M+ nodes at 30+ FPS. Built in 5 days with AI-assisted development. ML-powered antibody engineering and drug discovery features.
- Website: https://taxonsphere.com
- Page: https://www.paulmangiamele.com/work/phylogenetic-visualization

### Roche NAVIFY Mutation Profiler (2019)
Clinical genomics platform transforming cancer detection through liquid biopsy and next-generation sequencing. Deployed to 500+ clinical laboratories worldwide.
- Page: https://www.paulmangiamele.com/work/navify-mutation-caller

## Research Focus

Why are experts skeptical of computational tools and AI, and what does it take to earn their trust? This question drives everything — from clinical genomics to drug discovery to autonomous vehicles. Scientists retain trust when AI cites evidence with chain of custody, surfaces uncertainty explicitly, and provides controls at each decision point.

## Contact

- Portfolio: https://www.paulmangiamele.com
- LinkedIn: https://linkedin.com/in/paulmm
- GitHub: https://github.com/paulmm
- X: https://x.com/paul_in_ux