AI Systems Architect | Remote

About TubeScience Labs

TubeScience Labs is the applied-AI team inside TubeScience — the largest performance-video company in paid social. TubeScience is Meta's largest creative partner and AppLovin's #1 creative partner, producing 8,000+ original ads every month from a 100,000 sq ft Los Angeles studio, backed by a library of 1.6 million+ performance ads and $2B in annual managed ad spend. That makes one of the richest first-party creative-performance datasets anywhere.

Labs turns that data — and the playbook behind billions in spend — into frontier AI tools that actually ship. Our products run in production against real creative, real deadlines, and real budgets every day. The tools that graduate internally become products we ship to external clients.

 

About the Role

TubeScience is hiring an AI Systems Architect to serve as the technical lead for the infrastructure at the core of our production AI platform — the media pipelines, LLM routing layers, distributed data systems, agentic AI systems, and observability tooling that everything else depends on. You'll own the technical direction for how these systems are designed, working closely with product and engineering across the org, and then you'll build them.

This is a deeply hands-on, greenfield role. You'll write code, set architectural direction, and make the hardest technical calls — but you won't be executing someone else's roadmap. You'll be one of a small number of engineers who decides not just how to build, but what to build, as the platform and the AI systems underneath it evolve in tandem. The decisions you make early will be load-bearing for years.

We care as much about craft as we do about capability. The right person obsesses over how systems behave under real load — reliability, observability, the failure modes that only show up in production — and knows how to put the foundations in place so a fast-moving team can keep that bar high.

This role is fully remote.

 

In this role, you will:

• Own the infrastructure core: Set the technical strategy for the media pipelines, LLM routing, distributed data systems, and observability tooling that the entire platform runs on — and architect them from first principles.
• Design AI agent systems: Build AI agents and multi-agent systems from the ground up — orchestration, tool-calling, MCP servers, A2A dispatch, and the permissions and observability infrastructure that makes them reliable in production.
• Go deep on distributed systems: Push into the hard parts — data flow, fault boundaries, consistency trade-offs, performance under load — to build infrastructure that holds up when it matters.
• Raise the bar on reliability: Design for observability from day one — tracing, metrics, structured logging — and build the quality systems that let a fast-moving team move without breaking things.
• Shape the roadmap: Bring an engineering point of view to platform strategy — what we should build, in what order, and why.
  
  

You might be a good fit if you:

• Have set technical direction across teams, and can point to specific systems you architected and operated in production, and the moments where you brought people along.
• Are deeply expert in distributed systems including the layers below the framework — data flow, fault tolerance, consistency, and performance under real load.
• Design AI agents and agentic systems: orchestration, tool-calling, multi-agent coordination, and the infrastructure that makes them observable and production-ready.
• Have built large-scale platforms from scratch — making the foundational decisions and living with them — not just joined a mature codebase.
• Design clean, durable APIs across REST, gRPC, event-driven, and schema-based systems, and are fluent with MCP, A2A, schema registries, and tool-calling normalization.
• Know media pipelines deeply — transcoding, ffmpeg-class tooling, cloud storage, large-object processing — with strong hands-on depth across PostgreSQL, OpenSearch, Redis, Docker, Kubernetes, and Terraform or Pulumi.
• Are the person teams want in the room when a hard architectural call has to be made.

 

Strong candidates may also have:

• Production experience at the intersection of AI and infrastructure — LLM APIs, multi-provider routing and abstraction, schema normalization, structured output, semantic caching, and prompt engineering at the infrastructure layer.
• Experience building the plumbing that makes agentic AI systems reliable and observable at scale — in production, not just in demos.
• A track record of seeing a multi-quarter technical strategy through from proposal to outcome.

 

How we're different

Most AI tools never leave the demo. Ours can't hide — they're used by expert teams who notice every miss, against a real P&L, all day, every day. We're a small, senior group of scientists and veteran engineers from leading tech and consumer companies, and we hire deliberately for people who want hard problems, real users from day one, and room to own things end to end.

We work in tight loops: an idea can go from a prototype to a product in front of operators and clients in days. We start with unique leverage — proprietary data and the industry-leading playbook behind billions in ad spend — build the tools in-house, prove them in production at massive scale, and then ship the winners beyond TubeScience. Velocity beats perfection until something earns the right to scale.

Come work with us!

TubeScience Labs is the applied-AI team inside TubeScience, headquartered in Los Angeles. You'll join a small, senior group with an outsized surface area: real autonomy, a one-of-a-kind dataset, and a direct line from your code to real customers.

We offer competitive compensation, comprehensive benefits, plus generous access to frontier and open-weight models for day-to-day work, prototyping, and evals. Most of all, we offer the rare thing in applied AI: hard problems, expert users from day one, and the room to take an idea from prototype to shipped product. Come build at the frontier with us!

We encourage you to apply even if you don't believe you meet every single qualification. Not all strong candidates will check every box as listed, and the best people we've hired rarely did. If this work excites you and you think you could do it well, we'd rather hear from you than have you rule yourself out — so please submit an application.