StartupBlink: case study

StartupBlink is a global startup-ecosystem intelligence platform. It ranks 300,000+ startups across 2,000+ cities in 118 countries, and 40+ government clients rely on it.

Everything a lead engineer does. I set the architecture and the standards, review the code, plan delivery with seven engineers and keep production healthy. I also own the tooling that makes the team faster, including how we use AI.

Here is the path a request takes.

• +The architecture of a Turborepo monorepo: TypeScript, Node.js, tRPC, Prisma, MySQL, Redis and React/Next.js.
• +The migration of legacy JavaScript into a TypeScript monorepo of 10+ packages and apps, with shared services reused across 5+ apps.
• +Operating and tuning the Cloudflare-fronted, Redis-cached setup.
• +CI/CD that ships 2+ production deploys a week.
• +Structured AI-assisted engineering for the team: coding agents, MCP integrations, custom skills and multi-agent workflows, all under human review.

1. One typed monorepo.Legacy JavaScript moved into a TypeScript Turborepo of 10+ packages and apps. Its shared services are reused across 5+ apps.
2. Cache before compute.Cloudflare in front, Redis behind: a ~70% cache-hit rate on ~18M requests a month.
3. Ship often.2+ production deploys a week, with deploy time cut by 50%.
4. AI-assisted, human-reviewed.Agents help with the work. People review everything that ships.

The results I can put numbers on. The engagement jump came after an end-to-end platform refactor that I led.