VP, Agent Engineering

Turn/River•San Francisco, CA

5d•Hybrid

About The Position

Turn/River is seeking a VP of Agent Engineering to unlock engineering capacity in our portfolio companies. You'll work directly with CTOs, CPOs, and engineering teams at our portfolio companies, building AI agents that automate recurring feature development. This is a hands-on building role where you will write agent code, design orchestration architectures, and partner with engineers daily.

Requirements

15+ years in software engineering, with production systems you personally designed, built, and shipped
3+ years building agentic AI systems in production - architecting orchestration, wiring tool use, debugging agent failures - not just directing others who do
Fluency in LLM orchestration patterns with strong convictions about agent reliability that come from lived experience, not just white-papers
Productive in a large, unfamiliar codebase within days/weeks
Experience in B2B or enterprise software environments, including comfort with on-prem, legacy architectures, and compliance constraints
Ability to walk a CTO through a technical strategy and back it up with a working demo
Effective across organizational boundaries - equally comfortable pairing with engineers and presenting to executive teams

Nice To Haves

Consulting, PE or VC portfolio operations, or multi-company technical leadership is a plus, but not required - we want hands-on operators

Responsibilities

Identify recurring, high-yield feature patterns in portfolio company backlogs that consume disproportionate engineering time and effort
Decompose those patterns into agent task hierarchies, orchestration flows, and validation loops
Prototype agents end-to-end yourself, proving the approach and value on real backlog items before involving other engineers
Use data and build working demos to earn CTO/CPO buy-in in order to dedicate the engineering team to agent development
Technically lead the agent team: review architecture, pair on hard problems, set quality bars for agent output
Instrument and track success: agent evaluation and accuracy, engineering capacity freed, roadmap velocity, defect rates
Apply learnings across the portfolio, sharpening the approach and playbook(s) with each engagement