Senior Applied ML Engineer, Evals & Data
Cardboard, Inc, Bengaluru
What you will do
- What "good" means for our agent, and datasets built from real usage.
- Offline and online evals: automated checks, model graders, and human review.
- Regression gates, and quality tracked alongside latency and cost.
- Which models the agent runs on.
What you need
- You've shipped an LLM or agent system that real customers used.
- You've built evals, datasets, or AI quality systems before.
- You can turn "the agent feels dumb" into a number, then move it.
- Strong in TypeScript or Python. No PhD needed.