Lead AI Native Engineer

Job description

Job responsibilities

• Build vertical agents. Own end-to-end agents for high-value workflows across research, software, hardware, data, and operations—from problem definition through production use.
• Establish the agent foundation. Build reusable primitives for models, tools, orchestration, context, memory, retrieval, permissions, human approval, and long-running execution.
• Create the context layer. Connect agents to trusted data through APIs, pipelines, MCP servers, and integrations with clear provenance and access control.
• Own evaluation and reliability. Build benchmarks, regression tests, tracing, monitoring, and failure-analysis loops across quality, latency, cost, security, and resilience.
• Advance frontier agent usage. Evaluate new models, coding agents, SDKs, and patterns, then turn useful capabilities into maintainable systems rather than demos.
• Support strategic initiatives. Help company leadership apply agents and analytical systems to market and customer intelligence, partnerships, fundraising, diligence, scenario analysis, and executive decisions.
• Provide technical leadership. Set architecture and engineering standards, review designs and code, and create reusable patterns for the technical team.

Job requirements

• 5+ years in software engineering, applied AI, ML systems, or a related field, with strong zero-to-one technical judgment.
• Experience at a frontier AI lab, leading AI company, or comparable team working at the edge of current model capabilities.
• A track record shipping production agentic systems that real users depend on—not only prompts, prototypes, or demos.
• Deep experience with frontier model APIs, tool use, orchestration, context engineering, retrieval, memory, and multi-step workflows.
• Experience building evaluations, regression tests, observability, and production failure-analysis loops for AI systems.
• Exceptional fluency with AI-native development workflows using Claude Code, Codex, Cursor, agent SDKs, or equivalent systems, with a rigorous understanding of where agents work and fail.
• Strong product judgment: able to turn an ambiguous decision or workflow into a useful, secure, measurable system.
• Requires 5 days/week in-office collaboration with the team.