Senior Software Engineer, Agent Platform (AI for the Planet)

The Allen Institute for Artificial Intelligence•Seattle, WA

4d•$146,880 - $220,320•Onsite

About The Position

We are building an agent platform for conservation groups, government agencies, and research organizations that want to run AI agents against their own data. They know their domain and they have the data. We give them the application infrastructure: the runtime, the APIs, the tooling, and the operational support to get an agent from a prototype to something they can rely on. You would build that foundation. That means the APIs those teams call, the SDK and CLI they install, the docs they read, the traces they use when something breaks, and the evaluation tooling that tells them whether their agent is any good. We want a developer outside our team to get from an idea to a running, observable agent without needing us on a call. You would also be one of those developers. We build and operate our own agents on this platform for real users, and you would spend real time doing that too. It is how we find the problems before our customers do. The work spans both sides of the platform. You would build the services that run agents, and you would build the interfaces other engineers use to reach them. You would work closely with the front-end team building the user-facing product on your APIs, and with product management on where platform investment pays off most. You would also spend time with customers directly, watching them build, hearing where they get stuck, and bringing that back into the platform.

Requirements

8+ years of professional software engineering experience.
Range across web services, distributed systems, data pipelines, and cloud tooling.
A track record of shipping products to external users and owning them in production.
Opinions about API design, and a preference for simple designs over clever ones.
Willingness to write documentation as part of the work rather than after it.
Fluency with modern AI tools in your own workflow.
A continuous learner who adapts quickly to the rapidly evolving landscape of LLMs and agent architectures.
Clear technical communication with engineers and non-engineers

Nice To Haves

Experience building developer platforms, SDKs, or public APIs for outside consumers.
Hands-on experience shipping LLM agents to production, including tool use, orchestration, sandboxing, and cost and latency tuning.
Experience building evaluation or experimentation systems for nondeterministic software.
Small or growth-stage company experience.
Public technical work such as open source contributions, writing, or side projects.
Willingness to travel on occasion.

Responsibilities

Design and evolve the platform APIs every agent depends on, covering registration, auth, secrets, deployment, runtime, events, discovery, and evaluation. Keep them stable, documented, and pleasant to use.
Build the observability layer that shows developers what their agents are actually doing: logs, conversation history, execution traces, metrics, and failure reporting.
Make evaluation a first-class part of the platform, so developers can define and run evals, compare versions, track quality over time, and see regressions early.
Build and operate the services underneath all of it: identity and authorization, runtime orchestration, deployment, event routing, and the agent registry.
Build agents yourself and run them in production, including prompt and tool design, quality and cost tuning, and keeping them healthy once real users depend on them.
Watch for patterns that keep coming up in customer work and turn them into real platform features.
Own the developer experience end to end: onboarding, quickstarts, SDKs, CLI tooling, example agents, API documentation, and a local development loop that works.