Forward Deployed Engineer

Title: Forward Deployed Engineer

Location: San Fracisco, CA

Duration: Full Time

Experience: 6-8 Years

MOI: Virtual

About the role:

Cognida builds production AI systems for enterprise clients: agentic orchestration, large-scale knowledge and data infrastructure, model serving and inference, evaluation systems, and the security and access-control layers that let any of it touch real data.

We're hiring a senior engineer with deep technical range across the AI systems stack: someone who reasons about the production behavior of LLM-based systems, distributed infrastructure, and data pipelines, not just how to call a model API.

What you'll do:

• Agentic and orchestration systems: multi-agent workflows, tool-use and agent-facing protocols (MCP or equivalent), state and memory management, tracing, replay, sandboxing.
• Large-scale data and knowledge infrastructure: entity/relationship graphs, structured extraction from unstructured data using LLMs in bounded, evaluable ways, incremental pipelines that avoid full reprocessing on every change.
• Model serving and inference: deployment, latency and cost optimization, reliability engineering for systems calling LLMs or client-hosted models at scale.
• Evaluation and cost infrastructure: gold sets, regression detection, per-operation cost meters.
• Access control and trust: permission and sensitivity enforcement built into the data layer, a single mutation/write gate every pipeline passes through.
• Core backend and data engineering: schema design, query planning, distributed systems fundamentals.

What we're looking for:

• 5+ years building production systems, with genuine depth in AI/ML systems: how LLM-based systems, distributed data infrastructure, and agentic pipelines fail in production (drift, cost blowup, latency cliffs, brittle orchestration), and how to design around it.
• Strong distributed systems and data engineering fundamentals: SQL/Postgres at a level where you reason about query plans and schema design under production constraints; comfort running these systems in production (monitoring, incident response, cost control).
• Direct production experience with at least two or three of: agentic/orchestration systems, knowledge/data infrastructure at scale, model serving and inference, evaluation systems, access-control and authorization design.
• Security-conscious by default.
• Comfort taking an architecture with open decisions and closing the gaps yourself.