Staff Software Engineer - Agentic Workflows

About the role

About us:

Causaly is redefining how humans acquire knowledge and develop insights in biomedicine. Our AI-powered platform enables researchers and decision-makers to discover and interpret evidence from millions of scientific publications, clinical trials, regulatory documents, and other complex data sources in minutes.

We are building the world’s most advanced biomedical knowledge platform, powered by a high-precision Knowledge Graph and GenAI capabilities. Our technology is already used by leading biopharmaceutical organizations to accelerate drug discovery, improve safety, and drive better decision-making.

Backed by top-tier investors including ICONIQ, Index Ventures, Pentech, and Marathon, we are scaling rapidly and expanding our product suite and market presence.

At a Glance

◆ What you'll build. The platform that runs our long-running orchestrated workflows in production. A single run can last hours, fan out across dozens of subagents, and has to produce evidence a scientist can audit.

◆ What you'll work in. Full-stack engineering, weighted towards backend and platform. TypeScript and Node.js, Restate for durable execution, PostgreSQL, Google Cloud and Azure.

◆ What we need from you. Significant backend depth in TypeScript, experience designing and operating distributed systems, and LLM-backed systems you have run in production.

What You'll Do

◆ Workflow orchestration. Design and evolve the workflow definition model: control flow, fan-out, conditional execution and data passing between steps, plus the static validation that rejects an invalid workflow before it runs.

◆ Durable execution. Build the engine that runs workflows, including checkpointing and resumption after failure, retry and abandonment policies, and idempotency guarantees.

◆ Monitoring, tracing and cost attribution. Instrument runs that cross several services and two languages, so that latency and token spend can be attributed to individual steps. These figures feed pricing decisions.

◆ LLM integration. Structured outputs, guardrails and retries, citation integrity across steps, and deciding where deterministic code should replace a model call.

◆ Architecture and technical leadership. Write the design proposal, break a large change into shippable increments, implement the hardest parts and own the result in production.

What We're Looking For

◆ Significant backend engineering experience in TypeScript and Node.js.

◆ Experience building LLM-backed systems that run in production, with informed views on how they fail.

◆ Experience designing and operating distributed systems.

◆ Practical experience of correctness under partial failure: idempotency, at-least-once delivery, safe retries, and backwards compatibility between services that deploy independently.

◆ Solid understanding of relational databases, including schema design and safe migration.