Arcadia
Principal Software Engineer - Agent AI Platform
$250K – $350Kposted range
Range as posted by the employer.
Posted 1 month ago · listed on Lever
Should you apply?
- Open
- 54 days
- Open a while. Either hard to fill or the process is slow. A direct note to the recruiter helps here.
- Open roles at this company
- 16
- A few open seats. Your application will get a real look.
- Salary
- Posted
- The employer put pay in writing. Anchor on it in the first conversation.
Days open comes from the posting date. Fill time and reposting rate come from this company's recent public postings, last refreshed May 20.
About the role
Why This Role Is Important to Arcadia
Teams across Arcadia already put agents to work enabling clients' clinical, analytics, and operations teams. We use them heavily for our own work, too. Much of the platform foundation is in place. Now it has to handle longer-running workflows, client-facing agents, and PHI at scale.
You'll join the team that owns the platform underneath these agent workflows. This isn't a solo mandate. Your influence comes from what you ship and what other teams build on top of it.
How We Build
Engineers at Arcadia own outcomes end to end: the problem, the design, the code, the rollout, and the result in production. We build alongside AI without handing it our judgment; our automation is governed and transparent, built to strengthen human expertise, not replace it. We have few layers, high agency, deep ownership. We adopt new tools and practices quickly, which lets us move faster than many startups while keeping the discipline a regulated environment demands.
Our agent platform sits on top of governed access to AWS-based petabyte-scale lakehouse data with open table formats, distributed Spark processing, and Kubernetes services. The current implementation has an OSS agent harness and a gateway/router across model providers, all built on a third-party agent framework.
What Success Looks Like
Arcadia product teams should be able to spin up an agent quickly, ship it with confidence, and evolve it over time safely and cost-effectively. The platform provides the building blocks: persistence, workflows, memory, and identity-aware, compliant connectivity to tools. The platform also takes care of the quality attributes: caching, compaction, evals, guardrails, observability, resiliency, and safe handling of PHI.
Teams adopt it because it's the fastest safe path, not because they're told to. You treat it as a product and can show the impact in numbers.
Next: customers bring their own agents to Arcadia, and eventually assemble them from our building blocks.