Staff Site Reliability Engineer — AI Platform

You'll need:

• 5+ years in SRE / platform / infrastructure engineering, including production ownership at significant scale.

• Hands-on experience operating LLM-backed systems in production: provider APIs (Bedrock, OpenAI, Anthropic, or similar), inference pipelines, self-hosted or managed model serving.

• Deep cloud-native background: AWS, Kubernetes, Terraform/IaC, CI/CD.

• Strong observability practice (Prometheus/Grafana, OpenTelemetry, or equivalent) and experience defining SLOs for non-deterministic systems.

• Proven cost-optimization work: you can show where you cut cloud or inference spend and how you made cost visible.

• Staff-level influence: you've set technical direction beyond your own team and brought others along

It would be great if you have:

• Experience building or operating an LLM gateway/proxy (e.g., LiteLLM, Kong AI Gateway, custom).

• Experience with GPU workload optimization, quantization, or serving frameworks (vLLM, TGI, Triton).

• Experience with eval pipelines and quality monitoring for LLM outputs.

We care deeply about your growth, well-being, and comfort:

• 🌍 Hybrid onboarding to start work remotely and relocation support for you and your family.

• 💙 Comprehensive health insurance for both you and your family.

• 📚 Professional development budget for conference tickets, online courses, and other relevant resources to help you grow.

• 🫶 Flexible benefits package: no one-size-fits-all perks here. You get a budget and you decide where it goes, from health and wellbeing to family and setting up your home office.

• 🪴 Hybrid work and generous, flexible time off - planned with your team, not rationed by a rigid quota.

• 🍽️ In-office perks, including free meals and snacks.

• 🤝 Company-funded sport activities, annual offsites and team-building events.

• 🤖 AI isn't a perk here - it's how we work. Claude, OpenAI, and more are on by default from day one, and we back teams in adopting whatever makes them faster.

Want to shape an AI platform's architecture from day one, instead of just maintaining what someone else built?

Manychat runs AI features for businesses worldwide, and we're hiring an SRE to own the reliability, performance, and cost of that AI infrastructure, and to raise the bar for how our whole engineering org builds on LLMs.

This isn't a classical SRE role with a bit of AI sprinkled on top. We need someone AI-native: you already understand how modern LLM systems behave in production, things like token throughput, provider rate limits, degraded model quality, and inference latency tails, and you treat all of that as a first-class reliability concern, not an afterthought.

Why this role is worth your time:

• The AI Platform is still young, so you'll shape its architecture, standards, and roadmap from the ground up.

• The stakes are real. AI features sit right in the critical path of customer-facing automation, so your work actually matters to the business, not just to a dashboard.

• You'll partner directly with the Head of Infrastructure, with real autonomy and visibility into how your decisions play out.

Sound like the kind of ownership you've been looking for?

• Own reliability and performance of our AI infrastructure: AI Gateway, inference services, and integrations with Amazon Bedrock, Azure OpenAI, and other LLM providers.

• Design and evolve the AI Gateway: routing, failover between providers, rate limiting, caching, and guardrails.

• Build observability for AI systems: latency/throughput/error SLOs per model and provider, token-level metrics, quality and drift signals.

• Drive cost optimization and FinOps for AI workloads: per-feature cost visibility, model right-sizing, caching strategies, provider mix.

• Run capacity planning and incident response for inference services; write and improve runbooks and postmortems.

• Scale AI expertise across the org: set standards, review designs, and coach teams shipping LLM-backed features.

• Green-field ownership: the AI Platform is young; you'll shape its architecture, standards, and roadmap.

• Real scale and real stakes: AI features sit in the critical path of customer-facing automation.

• Direct partnership with the Head of Infrastructure; high autonomy and visibility.