Responsibilities
- Backend services in NestJS: API design, request validation with Zod, event publishing through the canonical SDK, projection consumers, RLS-aware database access. You write the service, the tests, and the docs.
- The Next.js console: App Router, server and client components, the in-house design system, real-time UI patterns, the widget framework. You make complex platform state feel direct and obvious to operators.
- LLM integration: agent orchestration inside the workflow runtime, prompt engineering with version control, evaluation harnesses, model gateway usage, structured output validation, retry and fallback logic. You treat model calls as production code, not prompts in a notebook.
- Observability: OpenTelemetry traces, structured logs, meaningful metrics on every meaningful surface. If a customer reports a problem, your code should make finding it trivial.
- Engineering culture: you set the bar for engineers who join after you. The next hires learn from your PRs.
Requirements
- 7+ years of production TypeScript and Node. You write strict TypeScript by reflex; you don't reach for any.
- Production experience with NestJS or a comparable Node framework at the level where you understand the request lifecycle, dependency injection, and the trade-offs.
- Production experience with Next.js (App Router preferred) and modern React: server components, streaming, client/server boundary discipline.
- LLM-powered features shipped to production. You've dealt with prompt drift, eval design, hallucination handling, and the cost of getting any of those wrong.
- Postgres at a working level: migrations, index design, query plan analysis, trade-off reasoning.
- Event-driven architecture experience: pub/sub, idempotent consumers, eventual consistency, ordering, retries, dead-letter behavior.
- Strong testing instincts: unit, integration, end-to-end. You don't ship without tests and you care about the difference between tests that pass and tests that matter.
Nice to Have
- Drizzle ORM, Temporal, or Redpanda/Kafka in production.
- OpenTelemetry beyond surface-level — sampling strategy, semantic conventions, custom instrumentation.
- Design system contribution — component libraries, accessibility patterns, design token discipline.
- Agent frameworks (LangGraph, custom orchestration), eval tooling (Braintrust, LangSmith), or prompt versioning systems.
- Open-source contributions to the TS/Node/Next ecosystem.
Work Arrangement
Hybrid
Additional Information
- Applicants must be currently authorized to work in the United States on a full-time basis. The company does not anticipate providing sponsorship for employment visa status (e.g., H-1B, TN, etc.) for this individual position.
- The team is in office on Tuesdays, Wednesdays and Thursdays.
