Staff Software Engineer, Frontier Security Team
Core
Lead the design and development of Snowflake's Agentic Harness and agent evaluation platform for enterprise customers, ensuring high quality, security, and reliability of production-grade LLM applications and intelligent agents.
Role type
Staff Software Engineer (AI Infrastructure & Agent Systems)
Builds
Agentic Harness, agent evaluation platform, simulation/replay infrastructure, and multi-tenant model services
Domain
Enterprise AI, Large Language Models (LLMs), Agentic Workflows, Distributed Systems
Deliverable
production ML models | product features | infrastructure
Required skills
Technical leadership of complex production systems, LLM application and AI agent development, distributed systems architecture, agent runtime or workflow engine design, nondeterministic system evaluation, Python, systems/application languages (Java/Go/Rust/TypeScript), tool calling and prompt management, production observability, technical direction across teams, mentorship
Preferred skills
Evaluation/observability infrastructure for agentic systems, human-evaluation program design, multi-agent orchestration, synthetic task generation, vector search and retrieval systems, multi-tenant sensitive data operations, model fine-tuning and reinforcement learning
Technologies
Python, Java, Go, Rust, TypeScript, Kubernetes, SQL engines, cloud-native infrastructure
Responsibilities
Architect and build the Agentic Harness for multi-step AI workflows; Design stable interfaces for tool execution, state management, and permissions; Own agent quality end-to-end via evaluation harnesses and automated graders; Analyze production agent trajectories to identify and fix failures; Develop offline and online measurements for correctness, safety, and latency; Build simulation and replay infrastructure for large-scale experiments; Define technical direction and mentor engineers across multiple teams
Seniority
Staff, hands-on IC with technical leadership