Engineering case studies
The problem, the idea, how I did it and what changed, with the numbers.
Bringing server costs down, service by service
−75%idle AI podsOur platform was built for thousands of users and was serving 30 to 100 a day. I went through it service by service and cut what it reserved, while keeping live answers fast.
Jun – Oct 2026 · 5 monthsAI answers that are cheaper, better and configurable for every customer
$0.0012per answerI own the engine behind BHyve's AI answers, which searches a company's knowledge base and connected tools like SharePoint, Gmail and Outlook. This is how I made it cheaper per question, measurably more accurate, and configurable per customer without a fork.
Jun – Oct 2026 · 4 monthsMoving AI search from Postgres to LanceDB
5–6 msvector searchOur AI search ran inside Postgres, which scanned every chunk a user could see on every question. I moved it, in stages and without downtime, onto a dedicated vector store that keeps the same speed as the data grows and runs inside a customer's own cloud.
Jun 2025 – Sep 2026Document extraction, rebuilt and priced per page
40×faster Lite tierAI answers are only as good as the text we pull out of documents. I benchmarked the extraction engines, built a fusion pipeline that beats our old one on tables and headings, and turned it into three tiers each customer can pick by speed and cost.
Sep 2025 – Oct 2026From pentest report to fixes in a day
13 / 15findings fixedEnterprise customers buy security first. When a penetration test came back with 15 findings, I closed 13 of them the day after the test window ended. When a critical Next.js flaw was being used against our production website, I patched it while the attempts were still coming in.
May – Sep 2026Releases by tag, and a CI that costs a third
−71%CI minutesOne library change used to rebuild six apps. I redesigned how the monorepo builds and ships: merge to deploy, tag to release, and a CI that only builds what changed.
Apr – Oct 2026Leading a small team with reviews, a roadmap and AI pair-programmers
146PRs mergedOn a five-person team, leading means reviewing, unblocking and deciding what's next. I also built the tooling that lets each of us ship like a bigger team: AI agents with shared rules, and a loop from customer tickets straight to the board.
Jul 2025 – Oct 2026 · 15 monthsPlatform v2: one monorepo instead of thirteen repos
14 appsin one repoBHyve v1 was one frontend talking to twelve separate backend repos, each with its own chart and deploy. I started the new web app and became the largest contributor to Platform v2: one Nx monorepo with 14 apps, 24 shared libraries and one release process.
Nov 2024 – Feb 2026Owning the whole frontend for two years
97%of frontend commitsBefore Platform v2, I was effectively the only frontend engineer: the product, the marketing websites and everything customers clicked on. 671 of the 693 commits to the main app in that period were mine.