Human Delta is the AI reliability control plane for the autonomous enterprise. We run the knowledge AI agents answer from, and we grade every answer against it — so whether an agent is talking to a customer, an employee, or another agent, it's answering from the truth. We call that state Delta Zero: no gap between what your agents say and what's true.

The Role

We are hiring one title across multiple strengths. You will own large, critical parts of the system. We hire across a few different spikes, but the bar is exceptional individuals. Your spike may be:

Retrieval, embeddings and knowledge graphs.
Ingestion at scale.
Evals and applied ML.
Integrations
Platform, runtime and reliability.

What We’re Looking For

2-4 years of experience building and operating production software.
Production experience with TypeScript and Node.js or Bun.
Strong PostgreSQL knowledge and experience with Redis, job queues, background workers, or workflow systems.
Experience building APIs, integrations, webhooks, and data pipelines for production-grade, multi-tenant software.
Familiarity with observability, distributed systems, and agent evaluation concepts.
Strong ownership, product judgment, and willingness to work in person in San Francisco.

You do not need to have built a commercial evaluation platform before. You should, however, understand why evaluating nondeterministic agent systems is different from conventional unit testing and be excited to develop deeper expertise in the space.