REAL WORK,
REAL RESULTS
Selected consulting and product cases. Each one involved hands-on work — research, code, AI tooling, and strategic decisions backed by data.
Strategic Market Analysis for a HealthTech Startup
A HealthTech startup needed to validate their B2B market entry strategy. The founding team had strong domain expertise but lacked structured market evidence to choose a beachhead segment and prioritize their first product.
Built a custom multi-agent deep research system (4 AI agents with adversarial verification) to compress weeks of market analysis into days
Conducted full STEP analysis: 47 evidence items from 25+ sources, credibility score 8.0/10
Delivered 11-section strategic report: macro trends, competitive landscape, risk register with triggers, 3 development scenarios
Identified beachhead segment and V1 product with projected 240% ROI in first month
From Research to Deployed MVP in 2 Days
Explored a women's health niche for a consulting client. Went from zero knowledge to a deployed product with clinical scoring quiz, landing page, and analytics — in 48 hours.
Analyzed 22,700 user comments to extract real pain points and build a JTBD Canvas
Benchmarked 5 international competitors (Noom, Hims/Hers, Balance, Flo, Midi Health)
Integrated clinical assessment scales (MRS, Greene Climacteric) into a 16-question diagnostic quiz
Deployed full MVP in 1 day: 4 parallel AI agents (frontend, backend, content, QA), 1,200+ AI messages
AI Tooling as Product Infrastructure
Working with a 5-person team with varying AI literacy, I noticed the same research and analysis patterns repeating across projects. Instead of doing them manually each time, I packaged each into a reusable, open-source tool.
Multi-agent research with adversarial verification. Optimist, Pessimist, and Fact-Checker debate findings. Parallel sub-queries, bilingual.
Convert any research document into an interactive slide-based website with hover tooltips and animated transitions.
LLM-generated synthetic surveys with SSR scoring and ODI opportunity scoring. 90% correlation with real panels.
Takes a business description and target segment, generates a structured JTBD interview script ready for live interviews.
Feature Prioritization with Synthetic Research + ODI
A founder came to me with 15 feature ideas and zero clarity on which one to build first. Instead of spending 2 weeks and thousands of dollars on a research panel, I combined synthetic respondents with Tony Ulwick's Opportunity Scoring to get a ranked feature list in 30 minutes.
LLM-generated personas answer open-text questions about the product. Semantic Similarity Rating maps responses to Likert scale.
ODI scoring measures importance vs. satisfaction for each job-to-be-done. The gap = opportunity.
Result: ranked feature list. 3 features scored above 15 (build now), 2 below 5 (skip). The founder's favorite scored 7.
Have a similar challenge?
I help product teams find growth systematically. Let's talk about your product.