Staff AI Infrastructure Engineer

📋 Job Overview

Staff / Principal AI Infrastructure Engineer
Own the intelligence layer - the agent runtime, the inference substrate beneath it, and how that capability becomes product. Audit what exists; build what scales.

🏢 About Cassi

Cassi is a fast-growing startup building an intelligent home automation platform that enables property managers, service providers, and homeowners to easily maintain and operate a property (and more). We're a small team shipping real product daily - SOC2 compliant, event-driven, and built to scale.

The intelligence layer isn't a feature bolted onto that platform; it's increasingly how the platform works. Voice and chat agents that take real action on a property, retrieval over years of asset and service history, and ambient intelligence that surfaces what a property manager should look at before they think to ask.

🎯 The Role

We're looking for a Staff- or Principal-level engineer to own our intelligence layer end to end - the agent runtime, the multi-provider inference infrastructure underneath it, and the surfaces where that capability reaches real users.

Two things this role is not. It isn't a research role: everything you build ships to production and someone depends on it that week. And it isn't prompt-tuning: the hard problems here are systems problems - streaming, tool-call correctness, retrieval quality, authorization inside an agent loop, evaluation of non-deterministic behavior, and the unit economics of inference at scale.

It's also explicitly a productization role. Capability that only exists in a graph isn't worth much. You'll drive how intelligence shows up across customer-facing product and in the internal tooling our own team runs the business on.

✅ Key Responsibilities

• Architectural Audit & Scale: pressure-test existing design choices and own migrations
• Agent Runtime & Orchestration: own graph topology, tool-call correctness, and multi-turn state
• Inference Infrastructure & Model Routing: own provider-agnostic abstraction for streaming, failover, and routing
• Realtime Voice: manage live voice sessions, session lifecycle, and audio metrics
• Retrieval, Context & Memory: own embedding strategy, vector retrieval, and context assembly
• Ambient Intelligence & Evaluators: develop event-driven intelligence and evaluators
• Evaluation, Observability & Unit Economics: create eval harnesses and cost attribution systems
• Productization — In-Product and Internal: drive intelligence surfaces in products and internal tools

📌 Required Qualifications

• Staff- or Principal-level experience in systems engineering or infrastructure
• Deep understanding of distributed systems, streaming, and real-time processing
• Experience with multi-provider AI infrastructure and model routing
• Strong background in building and scaling production systems
• Experience with vector retrieval, RAG, and context assembly
• Ability to lead architectural decisions and drive productization

⭐ Desirable Experience

• Experience with voice and chat agent runtimes
• Background in ambient intelligence and event-driven systems
• Experience with evaluation harnesses and observability tools
• Experience with property management or home automation platforms

🎁 Benefits

Competitive salary and benefits package