As a Member of Technical Staff in Applied and Agentic AI, you are building the agentic layer of ai&, the frameworks, workflows, and tooling that make it possible for our internal teams and our customers to deploy agents on top of ai& infrastructure with as little friction as possible.
One of the most important things ai& is building is the ability for anyone to harness AI agents without needing to understand the complexity underneath. That means the agentic layer needs to be powerful enough for sophisticated engineering teams and accessible enough for customers who simply want agents that work. You are responsible for both. You will integrate and extend the best open-source agentic frameworks available, build custom workflows that demonstrate the full capability of ai& infrastructure, and continuously raise the bar on how easy and reliable agent deployment feels.
This role sits at the intersection of applied ML engineering and product. You will work closely with the model, and platform teams to make sure the agentic layer reflects the best of what ai& can do at every level of the stack. At the same time you will be close to customers, understanding how they want to deploy agents and removing every obstacle between them and a working system. You will also contribute to the evaluation and QA process for the models we ship, ensuring that what reaches customers is stable, capable, and continuously improving.
The ideal candidate has built real agentic systems before, cares deeply about developer experience, and takes personal pride in making complex technology feel simple. If you want to work on the part of the stack that people actually touch, this role is for you.
Responsibilities
- Agentic Framework Development Build and maintain the agentic frameworks that power agent deployment on ai& infrastructure. Integrate and extend open-source solutions including LangGraph, AutoGen, MCP, and others to create robust, production-ready tooling for internal teams and external customers.
- Customer-Facing Deployment Tooling Make deploying agents on ai& as simple as possible. Build the interfaces, workflows, and abstractions that let customers go from idea to running agent with minimal friction.
- Custom Agentic Workflows Design and build custom agentic workflows on top of the ai& inference product that demonstrate real capability and serve as the foundation for customer deployments.
- Continuous Evaluation & Model QA Own the evaluation pipeline for inference models. Design evals that reflect real-world usage, catch regressions, and provide the signal needed to make confident decisions about model publishing and deployment.
- Internal Enablement Build the agentic tooling and workflows that our own engineering teams use every day. Make ai& the best possible platform to build on, starting from the inside.
Requirements
- Agentic Framework Experience You have built production agentic systems. You understand orchestration, tool use, memory, multi-step planning, and how to make these systems reliable at scale.
- Applied ML Fluency You know how to work with models in practice — prompting, chaining, evaluating, and iterating. You understand what it takes to turn a capable model into a dependable product.
- Evaluation & Testing Rigor You have designed and maintained eval pipelines for AI systems and you know that evals are only useful if they reflect what users actually care about.
- Developer Experience Instinct You think hard about what it feels like to build on a platform. You reduce complexity, write clean interfaces, and make the right patterns easy to follow.
- Full-Stack Comfort Comfortable building across backend services and APIs. You can own a workflow end to end.
- Relevant Tooling Strong command of Python. Familiarity with agentic frameworks (LangGraph, AutoGen, or similar), evaluation tooling, and inference APIs.
- Great Team Spirit A mission-driven approach to engineering, valuing clear communication, hands-on execution, and collective success over individual silos.
