Building Better AI Systems Through Evaluation-Driven Development
Andrew Ng's agentic AI framework applied to internal tools, customer-facing products, and continuous improvement
Read →Nate Steinrueck
UX researcher who builds
Mixed-methods UX researcher in a highly specialized, cross-functional role that combines research, systems thinking, and building AI systems.
Leads research and implementation efforts at IQVIA to leverage AI in internal tools and workflows. The work starts with in-depth, detailed analysis of complex work and systems to find where AI can best streamline existing workflows, then carries those findings through discovery and proof-of-concept builds that move into pilot programs.
Also builds internal tools to improve how research gets done, such as a synthesis pipeline that tags interview transcripts and drafts reports, and an agentic workflow for building research tools and prototypes with Claude and Cursor.
Thoughtful and rigorous about AI usage: benchmarks AI output against manual analysis, uses structured outputs so results stay consistent, and reviews everything before it informs a decision.
Companies
Selected Work
Research I've led and tools I've built, usually both at once.
Medicare Enrollment
Doubled digital enrollments. Led 20+ studies as sole researcher embedded with product and engineering.
Google Planning Tool
14+ studies for AI-powered space planning. Unblocked 50% of target users, informed 20+ feature improvements.
Research Copilot
Multi-agent AI for research synthesis. Python, FastAPI, React, GPT-4, FAISS.
AEC Hub
AI research, proposal software, and a firm AI workspace for AEC firms: aechub.org, proposals.aechub.org, and platform.aechub.org.
Articles
How I think about research, AI, and what's next for both.
AI-Augmented UX Research
Turning research insights into living organizational knowledge systems
Read →AI-Augmented Workflows for UX Research
How I cut research turnaround from 3 weeks to 1 week using AI
Read on Medium →About
About
I'm a UX researcher who builds end-to-end.
Most researchers hand off findings and hope something ships. I lead the research (50+ studies across healthcare, enterprise software, and design technology), then I build. Full-stack applications, AI agents, data pipelines, production features. Not prototypes-to-prove-a-point. Working software that solves the problem the research identified.
I got here through an unusual path. I started in architecture: designing buildings, managing construction projects in Central America, teaching computational design. That background gave me a bias toward making things, not just analyzing them. When I moved into product research, I brought the builder's instinct with me.
The tools I work with reflect that range: Python, React, Next.js on the engineering side. FastAPI, GPT-4, Claude, FAISS, CoreML on the AI side. Dovetail, Maze, Mixpanel, and good old spreadsheets on the research side. I don't think of these as separate skill sets. They're how I close the gap between insight and implementation.
Based in Atlanta. Currently at IQVIA. I do my best work on teams where research and engineering aren't separate conversations.
Say Hello
Say Hello
I like conversations about research, AI, and teams where understanding users and building intelligent systems are the same job.