Skip to content
Vivek Tumrukota

Playbook

How I work.

Principles I’ve relied on, from a nine-person startup to a 70-engineer org.

  1. 01

    Own the outcome

    Done means shipped and working. Whatever my title, I stay with the work until it’s in customers’ hands and holding up.

  2. 02

    Measure before debating

    At WorkSpan, Sprig and Trustero I put analytics and SLAs in place first, so roadmap discussions started from data.

  3. 03

    Judge AI by what ships

    Usage numbers say little on their own. What matters is whether teams ship more, and ship better, so that’s how I pick tools.

  4. 04

    Test AI like any other feature

    Every AI feature and agent workflow gets a test set and a quality bar before it reaches customers.

  5. 05

    Let more people ship

    With the right guardrails, product managers, designers and support engineers can make real changes, which frees engineers for harder problems.

  6. 06

    Small teams, short loops

    A few senior people shipping often, close to customers, beat a big team waiting on process. I set that up first, then scale it.

Tools and practices

What I use day to day, from the codebase and CI to models, evaluation and team design.

Agentic development
  • Cursor
  • Cloud agents & automations
  • MCP
  • AI code review
  • Agent test infrastructure
Models & evals
  • Anthropic
  • OpenAI
  • Braintrust
  • LLM development practices
Hands-on engineering
  • TypeScript
  • React
  • Go
  • Node
  • GraphQL / Relay
  • Ruby on Rails
  • Kotlin
  • PostgreSQL
  • Kafka
Platform & quality
  • AWS
  • GCP
  • Kubernetes
  • Terraform
  • Honeycomb
  • Datadog
  • Playwright
  • LaunchDarkly
Leadership
  • AI guilds & adoption playbooks
  • Hiring rubrics
  • Enterprise accounts
  • Cross-functional programs
  • Board & exec communication