AI Product Manager

Company Description CareerPassport is building evidence-led professional identity infrastructure designed for the AI era. The company focuses on helping knowledge workers showcase their real, lived experience and capabilities, beyond traditional resumes and profiles. By enabling talent evaluation based on concrete evidence rather than claims, CareerPassport supports more accurate and fair hiring decisions. The platform empowers professionals to present “the real you” and helps organizations hire for what is demonstrably real. Applicants joining CareerPassport will contribute to redefining how professional identity and talent assessment work in an AI-driven world.
Roles and Responsibilities
• Build and operate services — Design, ship, and run Node.js/TypeScript services that sit behind the GraphQL federation. Own a slice of the ~40-service estate end-to-end: API design, schema, persistence, tests, deployment, on-call.
• GraphQL federation subgraphs — Add types, resolvers, and capabilities to the federated Platform API. Keep subgraph boundaries clean; respect RBAC at the federation enforcement layer.
• Commerce backbone — Build the subscription, invoicing, overage, dispute, and payout flows that integrate with Stripe and PayPal webhooks and the billing state machine. Make sure metering, billing, and payouts stay accurate, auditable, and replayable.
• Proxy control plane — Build the management APIs and background workers that configure the runtime proxy: API keys, plans, rate-limit policies, quota caches, abuse signals.
• Event-driven systems — Produce and consume Kafka and Kinesis events safely: schema evolution, idempotency, retries, dead-letter handling. Operate Lambda consumers where they're the right tool.
• Data services — Help own the Node.js Analytics Service and Quota Service that front ClickHouse and Snowflake. Keep latency low, queries cheap, and authorisation correct.
• Performance and reliability — Diagnose hot paths, tune queries, add caches where they matter, and remove caches that have become liabilities. Write the SLOs and the alerts that defend them.
• Security isolation — Implement OAuth2 / JWT / mTLS where they belong; respect least-privilege; design APIs that fail closed.
• On-call and incident response — Take part in the rotation. Lead investigations across services when something rooted in your area breaks.
• LLM-assisted engineering — Use LLM-based coding and agentic tooling to evolve how features are designed, reviewed, and tested.
What Success Looks Like
• Services you own meet their SLOs, deploy multiple times a day, and rarely wake anyone up.
• Billing, subscription, overage, and payout flows you ship are accurate to the cent, replayable, and trusted by both providers and consumers.
• GraphQL subgraphs you maintain stay fast, well-typed, and pleasant to extend.
• Event pipelines you produce and consume are idempotent, observable, and survive replay and back-pressure.
• The first generation of agent-native primitives you help build land in production and start carrying real agent traffic.
What You Bring
• 5–8 years of backend engineering experience, with non-trivial work on production services that real customers depended on.
• Strong TypeScript / Node.js in a production service environment; comfort with async patterns, streams, and back-pressure.
• Solid API design — REST and GraphQL (ideally federation: Apollo Federation, Mercurius, or similar). Schema design, persisted queries, performance tuning.
• Production fluency with SQL — MySQL and/or Postgres. Bonus for Redis or ClickHouse.
• Experience with event-driven systems — Kafka or similar. You understand idempotency, exactly-once myths, and consumer lag.
• Hands-on AWS (EKS, MSK, Kinesis, DynamoDB, Lambda, DMS) and Kubernetes. Comfortable reading and writing Terraform.
• Modern engineering practice: observability, feature flags, CI/CD, and safe rollouts.
• Care for correctness on the money path — you have shipped billing, metering, or payment code before and have the scars to show for it.
• Bonus: experience with API gateways, rate limiting, quota enforcement, usage metering, agent protocols (MCP, A2A, x402), or LLM-based developer tooling.
• You write small, reviewable PRs. You add the test before you fix the bug. You would rather ship a smaller thing well than a bigger thing badly.

Work arrangement
No