Enterprise Solutions Engineer
Enterprise solutions engineer. 300+ demonstrations, deals from $250K to $1.5M, C-level and investor rooms.
An evidence scoring system for revenue teams. Corpus through interface, built around a single constraint: it does not produce a number it cannot support.
Nobody asked me to build this. I wanted to know whether a scoring system could refuse to answer, and whether buyers would trust it more for refusing.
A language model makes bounded per-record judgments. Does this record support or threaten this objective, and how strongly. One record, one question, one answer.
Every number is computed in deterministic code. The model never aggregates, never weights, never returns a score. Scoring is arithmetic over judgments, not a judgment about arithmetic.
Quoted evidence is matched verbatim against its source record. A span that cannot be found is discarded rather than paraphrased into something defensible-looking.
Below a coverage floor the system publishes nothing. Insufficient evidence is a supported output. A plausible number derived from two weak records is not.
All figures from the September 2026 run · corpus is synthetic and deterministically generated · longswell is a fictional vendor
Built with Postgres, Node, TypeScript, React and Docker Compose, scored through the OpenAI API. I specified the architecture and the scoring rules, drove the implementation with AI coding tools, and operated and debugged the pipeline when it broke.
Every role, technology, education and certification. Readable in the browser, no download required.
PDF Download a copyThe same content as a two page PDF, for filing or forwarding to a hiring team.
Profile LinkedInWork history, recommendations and the usual professional surface area.
Expanded role within the Product organization, continuing all customer-facing pre-sales responsibilities while formalizing the bridge between field solutions work and product strategy.
Rather than read the cards above, ask it something. Answers are generated from my résumé and the contents of this site.
Generated answers · verify anything that matters against the résumé
Reading an unfamiliar codebase What a schema's migration history tells you about the product that was actually built, why uniform error handling is a tell, where the real intellectual property concentrates, and which parts of a build collapse under AI assistance while others stubbornly refuse to.