Member of Technical Staff, DevOps

Member of Technical Staff, DevOps

Department: Engineering

Employment Type: Full Time

Location: San Francisco

Description Reactor is building the inference platform for real-time world models. We run the largest and fastest catalog of world models on the market (15+ and growing), including LingBot World 2, LTX 2.5 Realtime, and HappyOyster. Our founders, Alberto Taiuti (CEO) and Bryce Schmidtchen (CTO), met as technical leads on the early Apple Vision Pro team. Alberto went on to co-found Luma AI, now a $4B company. We've raised $59M, led by Lightspeed, with Amplify Partners, WndrCo, and NVIDIA's NVentures participating. Jeffrey Katzenberg, cofounder of Dreamworks Animation, sits on our board of directors.

We're looking for a DevOps / SRE engineer to own the reliability, delivery, and observability of our AI platform. You'll be the person who ensures models get from a developer's branch to production without anyone losing sleep — and when something does go wrong at 2am, you'll be the one who knows where to look.

We run production across multiple Kubernetes clusters, cloud providers, and regions. Our deployment pipeline is fully automated through CI/CD and GitOps, our infrastructure is managed as code, and our observability stack gives us full visibility across every service and GPU workload. This role is about making all of that faster, more reliable, and easier to operate as we scale.

What You'll Do

Own and evolve our CI/CD pipelines: dynamic pipeline generation across a monorepo of Go services, Python model containers, and Helm charts

Operate and improve our GitOps deployment lifecycle: Helm releases, Kustomizations, and image automation across multiple clusters

Build and maintain our observability stack: distributed tracing, metrics, dashboards, and alerting across all services and GPU workloads

Define and track SLOs for core platform services, including session latency, model cold start time, and streaming reliability

Run incident response: triage production issues, write postmortems, build runbooks, and drive reliability improvements

Manage infrastructure-as-code across multiple cloud providers and regions: plan/apply workflows, state management, drift detection

Operate secret management: encrypted secrets, external secret syncing, certificate automation

Improve deployment safety: canary rollouts, health checks, startup probes, rollback automation

Manage authentication infrastructure: OIDC federation for CI, workload identity for cloud services, cross-cloud credential management

Participate in on-call rotation and build the tooling that makes on-call less painful

What We're Looking For

You've run production Kubernetes clusters and been on-call for them. You've debugged node scheduling failures, OOM kills, and mysterious pod evictions at 3am

Strong CI/CD experience: you've built and maintained pipelines for monorepos, not just single-service repos

GitOps experience: you understand reconciliation loops, drift detection, and why image automation matters

Infrastructure-as-code fluency with Terraform or similar across multiple environments and cloud accounts

You know observability beyond just "set up dashboards". You've defined SLOs, built alerting that doesn't page on noise, and used traces to debug cross-service latency issues.

Comfortable with secret management patterns (KMS, encrypted configs, external secret operators). You've thought about credential rotation and zero-trust.

Incident response experience: you've triaged production outages, written postmortems that actually led to improvements, and built runbooks that other engineers could follow

You write code, not just YAML. Proficiency in Go, Python, or Bash for building tooling, automation, and pipeline scripts

Nice to Have

Experience with GPU workloads on Kubernetes: device plugins, GPU-aware scheduling, GPU monitoring

Multi-cloud operations beyond a single provider

Real-time or streaming workloads: low-latency systems where p99 matters more than average

Experience with Helm chart authoring and managing complex value layering across environments

Familiarity with real-time media or relay infrastructure

FinOps experience: GPU cost optimization, spot/preemptible instance management

What We're Not Looking For

Engineers who treat infrastructure-as-code as "click around in the console and import later"

SREs who've only monitored systems but never built the deployment pipelines that ship to them

Candidates whose CI/CD experience is limited to GitHub Actions for a single-service repo

People who write alerts that fire every day and then get ignored

Logistics We are based in-person in San Francisco. We are also hiring for this role in Europe for on-call coverage and timezone distribution.

Benefits

Competitive salary and meaningful early equity

Visa sponsorship and relocation support

Generous health, dental, and vision coverage