Backend / Infra Engineer

Lemma offers an AI platform that automates prompt optimization and continuous learning for model improvement.

Location: San Francisco, CA
Company size: 2–10
Posted: 3w ago
Via: Yc

Section II · Full ProfileFree with an account

01Comp band & equity packageLocked
02Seniority & experience requirementsLocked
03Interview process & rubricLocked
04Hiring manager & team contextLocked
05Growth trajectory in this roleLocked
06Offer & decision timelineLocked

Create free account Sign in

Free account · no card · 2 minutes

Job listing · Updated 21 AUG 2026

Backend / Infra Engineer

Lemma· San Francisco, CA, USView company profile

Job title: Backend / Infra Engineer
Job location: San Francisco, CA, US
Job listing URL: https://www.ycombinator.com/companies/uselemma/jobs/NIopCzH-backend-infra-engineer
Job description: Lemma reads all of your agent conversations, finds the failures nobody knew to look for, and opens the PR that fixes them. ### Why this role exists We sell reliability tooling. There is no version of this company where our own pipeline is the flaky part. Right now it works because it's small and we're watching it. This role is about making it work when it's large and nobody is watching it. The interesting problems here aren't modeling problems. Agent runs go for hours, call thousands of tools, and produce deeply nested spans that look completely different at every customer. We read all of production, not a sample, and we surface the anomalous fraction of a percent without anyone telling us what anomalous means. Doing that accurately is hard. Doing it at a cost per event that doesn't eat the business is the actual job. Then the part that decides whether we have a company: reproducing a failure we saw once, proving it's real and not variance, and being right enough that a team lets us open PRs against their repo. Being confidently wrong once costs more trust than being right fifty times earns. ### What you'll do * Own ingest: schema, throughput, cost per event, and the long tail of customers whose instrumentation is a mess * Build the detection pipeline that runs unsupervised across 100% of production data * Make the fix loop trustworthy. Reproduction, verification, and the guardrails that keep a bad PR from ever reaching a customer's repo * Own the deployment story for customers who won't send data outside their VPC. This is a live sales blocker, and solving it opens doors * Set the on-call, testing, and observability standards for our own systems, since nobody has yet ### What we're looking for * 4+ years on backend or data infrastructure, with something high-volume in your history you can talk about in real detail * Strong TypeScript. The whole codebase is TypeScript, including the core service and workflows * Comfort with columnar stores and the economics of storing a lot of events cheaply. We run ClickHouse, Quickwit, and Qdrant, and you'll have opinions about all three * Pragmatism about scale. We need architecture that survives 100x, built by someone who won't build for 100x on day one * Bonus: anomaly detection work where the ground truth was genuinely unknown * Bonus: you've built the enterprise deployment path before. Self-hosted, VPC, air-gapped We move fast on hiring. Target is an offer within two weeks of first contact. Onsite in San Francisco. We sponsor visas.

View job listing ↗

The Saturday Briefing