Staff Software Engineer (LLM Infra)

Software Engineer - Distributed Systems & Agentic AI Infrastructure

The Opportunity

We're partnering with a startup being incubated inside a top-tier venture firm.

They're not building another AI wrapper. They're building the hard infrastructure underneath agentic AI. The runtime that agents actually execute on.

The founding team is small, the technical direction is still being shaped, and the engineers who join now will define the architecture.

What You'd Be Building

• Reliable agent orchestration - fault-tolerant scheduling and execution of multi-step, multi-agent workflows at production scale
• Stateful multi-agent coordination - consensus, conflict resolution, and shared state across autonomous agents operating in parallel
• Low-latency tool execution - sub-second invocation layers that let agents interact with external systems without becoming the bottleneck
• Durable memory at scale - persistent, queryable memory backends that give agents long-horizon context without blowing up latency or cost

What We're Looking For

• Deep distributed systems experience - you've built, operated, or scaled systems at growing companies working with vast volumes of data.
• Production-grade engineering - replication, consistency, failure recovery, and state management aren't abstract concepts to you - you've shipped them
• Interest in the agentic AI stack - you don't need to be an ML researcher, but you see the infrastructure gaps in how LLM-powered agents are built today
• Early-stage appetite - you want to be in the room when the architecture is still on a whiteboard, with the backing of a world-class VC behind you
• Nice to have: Experience at the intersection of databases/storage systems and AI infrastructure (vector stores, retrieval systems, context management at scale).

Why This Role

The "agentic AI" wave is real, but the infrastructure is immature. Most agent frameworks today are single-threaded, stateless, and fragile. This company believes the unlock is at the systems layer - and they're hiring the distributed systems engineers who can build it. With Tier-1 VC incubation, you get the resources and network of a top firm with the autonomy and upside of a founding team.