Millee is an AI recruitment coach for recruiters and hiring managers, productizing the strongest recruitment techniques in the world, to make expert interviewing accessible to all.
What you'll be doing
You'll be one of Millee's first engineers. You'll own production end to end: real-time products that sit in on phone and video calls, agents that work insider recruiting CRMs and inboxes, and the platform underneath both. We're a small team that ships every day. You'll talk to recruiters and hiring managers, and they'll use what you build that week. A lot of our code is written with AI agents, so reviewing, simplifying, and deleting code matters as much as writing it.
What you'll own
- The platform TypeScript services on Kubernetes. You keep them fast, cheap and up, and you have a strong opinion on what they should run on.
- Integrations Syncing recruiting CRMs, email and calendars through OAuth, webhooks and rate-limited APIs, into data the product can trust.
- AI in production Agents and model pipelines with structured outputs, retrieval and evals, tested and measured like any other code.
- Real time Streaming audio and transcription in a desktop app, on tight latency budgets.
Who you are
- Kubernetes in production You’ve run it yourself: Terraform, Helm, Docker and CI/CD on GitHub Actions. You can debug a crashing pod at 2am.
- Postgres Schema design, migrations on a live multi-tenant database, indexes, query plans, and backfills that don’t lock tables.
- Workflows and queues Redis-backed queues, scheduled jobs, retries and idempotency, so every job runs exactly once.
- Third-party APIs OAuth, webhooks, pagination, rate limits and partial failures, across many vendors at once.
- TypeScript End to end: Node and Bun services, Next.js, strict types and runtime schemas.
- Security OIDC, fine-grained authorization, tenant isolation and encrypted credentials. Candidate data is personal data.
- Code review You read a pull request, spot the duplicated abstraction, and cut it.
Strongly preferred
- LLM systems The AI SDK, model routing, structured outputs, offline evals and cost per run. Prompts treated as code, with tests.
- Desktop audio Electron, audio capture, voice activity detection, streaming speech-to-text, auto-update and code signing.
- Observability Sentry, Prometheus and alerts that page someone before a customer notices.
Benefits and Perks
- Location: Remote in the USA; LA or SF strongly preferred
- Equity: Founding-team equity and real ownership
- Compensation: $130k-$150k salary