Systems · Evaluation · Adoption
I build, evaluate, and help teams adopt practical AI workflows.
Working practice
Build · Evaluate · Enable
I focus on AI evaluation, agentic workflows, and practical AI enablement.
I map real work, build AI-assisted systems with prompts, agents, MCP tools, APIs, and automation, then test how they behave beyond the ideal demo.
I evaluate outputs for accuracy, instruction adherence, reasoning, hallucinations, clarity, and edge cases—then add human-review checkpoints, documentation, demos, and training so teams know when AI should automate, assist, or escalate.
How I work
From working prototype to working practice.
Build
Prompts, agents, MCP, APIs, and automation.
Evaluate
Accuracy, edge cases, safeguards, and human review.
Enable
Demos, playbooks, onboarding, and training.
Case studies
The problem. The decisions. The result.
Tradings.Guru
- Problem
- Stock research meant moving between market data, social discussion, and other sources. Collecting that information manually was repetitive and hard to keep current.
- Approach
- Bring sources together with APIs, MCP integrations, agent context, and a shared cache; check what each result actually represents.
- Result
- A deployed research platform that brings fragmented information into one interface, with more explicit sourcing and methodology in its options-flow tool.
StreetScope AI
- Problem
- An opportunity map can mistake missing business records for an underserved market. Useful recommendations need coverage and category quality behind the score.
- Approach
- Build a location-intelligence pipeline, simplify unhelpful scoring, and evaluate coverage and cost before expanding.
- Result
- A regional prototype with substantial Florida coverage and a reusable public scraper. The product is paused after nationwide collection and ongoing data costs proved difficult to justify.
Cash Carnage 2
- Problem
- Real-time co-op has to stay understandable under pressure: players need clear attack warnings, consistent controls, and a shared game state.
- Approach
- Pair AI-assisted implementation with regression tests, local four-player checks, and hands-on playtesting.
- Result
- A public, no-install co-op arena shooter with a playable campaign and a repeatable test-and-refine development loop.
Latest note
8 min read
Keep building AI. Keep humans accountable.
A response to Dario Amodei on human oversight, AI-written code, and who should be allowed to develop American AI models.
Contact