About the role
Build and operate reusable AI infrastructure services that enable teams to develop, deploy, and scale agentic AI and generative AI applications securely and reliably. This role focuses on building foundational AI platform capabilities, including LLMOps, observability, governance, guardrails, retrieval systems, orchestration frameworks, and runtime infrastructure for autonomous AI agents.
Responsibilities:
- Build AI infrastructure to support a team of agentic and GenAI app developers. This includes building systems for guardrails, observability, governance, LLMOps, agent-to-agent orchestration, traceability, and other such infra elements that are unique to AI systems. This includes the following:
- Build and maintain the foundational infrastructure for agentic AI systems, including orchestration layers, tool registries, and execution environments that support multi-step, autonomous agent workflows.
- Design reliable and observable agent runtimes, implementing tracing, logging, and replay tooling across long-running agent sessions, with support for interrupt/resume, sandboxing, and failure recovery (e. g., LangSmith, Arize, and custom OpenTelemetry pipelines).
- Own memory and context infrastructure: architect short- and long-term memory stores, manage context window strategies, and build retrieval systems that give agents reliable, low-latency access to the right information at the right time.
- Develop and operate tool/API integration layers to build secure, rate-limit-aware, and schema-validated connectors that agents use to interact with external systems, ensuring safe and auditable tool execution.
- Define guardrails and safety infrastructure for autonomous agents* and implement policy enforcement, output validation, human-in-the-loop checkpoints, and permission scoping to ensure agents operate within defined boundaries in production.
Requirements:
- 6-9 years of software engineering.
- 2-3 years developing ML/AI systems.
- Can independently build a retrieval pipeline, a simple eval framework, and a production-ready API.
- Competencies: Engineering fundamentals, learning agility, and a system reliability mindset.