The Mission

We’re looking for an enthusiastic and opinionated Principal AI Engineer to define how AI agents transform Engineering workflows across industries such as Manufacturing, Aerospace, and Semi-conductor. You'll be building the foundations that power next-generation simulation and design tools used by industry-leading engineering teams. Our platform allows Forward Deployed Engineers (FDEs) and customers to build and deploy deep learning surrogates that solve massive engineering challenges.

Your mission is to architect the Agentic stack within this wider ecosystem. You will build a production-grade platform that enables our Product teams, FDEs, and customers to compose advanced AI workflows safely, transparently, and reliably.

• You will set the strategic direction for our platform's critical infrastructure and lead key implementation efforts.
• You are someone with the scar tissue of running agents in production.
• You have strong, reasoned opinions on emerging open standards (such as MCP, A2A, and ACP) and deep expertise in complex architectural patterns like durable execution and agent memory.
• You are passionate about building a world-class developer experience that transforms complex research into robust, deployed engineering solutions.

Core Responsibilities

You will serve as the principal architect for our Agentic ecosystem, responsible for the high-level design choices that define how agents run at PhysicsX. You will cover topics such as:

• Agent Observability: Own the implementation to enforce deep tracing, granular cost tracking, and observability across the lifecycle.
• Agent Deployment: Deliver an intuitive deployment lifecycle which simplifies questions around auth, discoverability and resource requirements.
• Agent Sandboxing: Architect secure environments to isolate and allow agents to operate safely.
• Agent Evals: Implement a stack that empowers domain experts to efficiently annotate traces, turning qualitative feedback into quantitative, reproducible evaluation datasets.
• Agent Governance: Implement robust identity and access patterns to ensure agents can safely and correctly act on behalf of users in a regulated enterprise environment.

The Tech Stack

• Core Platform: Python (Primary), Go or TypeScript (Secondary), Kubernetes, Docker, Terraform.
• Agentic Infrastructure: LangGraph/LangChain, Temporal (Durable Execution), Vector DBs (Pinecone/Weaviate).
• Observability & Evals: OTel, LangSmith, Arize, Braintrust.

Who You Are

• An Architect at Heart: You have strong, reasoned opinions on Durable Execution vs. Standard Async, Vector Search vs. Keyword Search, and Prompt Management strategies.
• Platform-First: You care deeply about the Developer Experience (DevEx) of the Customers, FDEs consuming your platform. You build tools that make the "right way" the "easy way."
• Security-Minded: You understand the risks of allowing LLMs to execute code and access data, and you know how to mitigate them via rigid sandboxing and permissioning.

Qualifications

• Platform & Backend Foundations:
  • 4+ years of experience in Platform Engineering, Backend, or SRE.
  • Strong proficiency in Python/Go, Kubernetes, Docker, and IaC (Terraform).
• Agentic & AI Engineering:
  • Production experience designing Agentic architectures (chains, tools, memory).
  • Familiarity with Agentic frameworks (LangGraph, PydanticAI) and patterns like durable execution.
  • Understanding of LLM-specific lifecycle issues: non-determinism, systematic evals, and token-based cost tracking.

Bonus Points

• Background in Engineering workflows or simulation platforms.
• Experience building internal developer platforms (IDPs).