ZON builds AI that makes AI better. We are an IP-centered company whose core technologies accelerate and enhance foundational commercial models through recursive, autonomously adaptive processes, licensed and applied with data-aware, AI-ready partners. Our flagship product, ZON Radiance, optimizes energy use inside data centers so operators get more compute from the power they already have.
The role
The work is not a chatbot wrapper. It is production systems other people have to live with: operators, customers, and the engineers who inherit them.
At ZON we use AI to create products quickly. Your job is to take what we build and push it to production, and to define what a production-level artifact means here.
You will design, deploy, operate and evolve production-grade AI agents, developer harnesses, multi-model orchestration systems and self-improving AI. The hard part is the system around the model: context and memory, tools and permissions, durable execution, evaluation, cost, failure recovery, human review, and the loops that let the system improve without becoming unreviewable.
AI-supported and AI-driven development is our primary way of building: coding agents, harnesses, skills, evals and review loops. You still read, shape and take responsibility for everything that ships.
This role is remote, anywhere in the United States.
What you will work on
1. Production readiness
- Take AI-built products from working prototype to production across ZON's products.
- Define and apply ZON's bar for a production-ready artifact: versioning, review, testing, observability, rollback.
- Treat prompts, skills, agent graphs and workflows as production software, with canaries, kill switches and audit logs.
2. Agents and harnesses
- Design multi-step, tool-using and multi-agent workflows (planner/executor, hierarchical, human-in-the-loop) with explicit state, retries and safe fallbacks.
- Build production harnesses: tool dispatch, session and state management, context as a managed resource, memory, compaction and recovery on long-running work.
- Design self-improving loops: production traces, eval suites, prompt/skill/tool updates, regression gates, rollout, with a human in the loop on high-impact actions.
3. Orchestration and durable execution
- Run durable execution on Temporal: retries, timeouts, workflow versioning and recovery across failures.
- Route across commercial and open-source models for quality, latency, cost, capability and policy.
- Integrate tools and internal systems through function calling, MCP, A2A, hooks, APIs, retrieval and structured outputs.
4. Operations, evaluation and safety
- Stand up evaluation as a first-class system: golden datasets, trajectory evals, automated graders where valid, human review where not.
- Instrument traces, token and cost accounting, tool-error rates, loop detection and drift.
What you need to have done
Core
- Shipped software other people depend on, including at least one LLM- or agent-backed system past the demo stage.
- Hands-on durable-execution experience (Temporal or similar), including failure recovery on long-running work.
- Can design an agent or harness as a system: tools, state, context, evaluation, durable execution, and what happens when a step fails.
- AI-driven development is how you build. You use coding agents, skills and eval loops as the main path from spec to shipped change, and you own the result.
- A real eval habit: you have caught a silent regression, written a grader, or refused to ship on a good vibe.
- Broad, current familiarity with commercial and open-source AI tools and frameworks.
Useful, not required
- Experience working inside a data center, with GPU virtualization, infrastructure management and machine learning
- Ability to work with diverse DevOps and facilities management groups
- Multi-model routing and cost/quality/policy tradeoffs
- CI/CD, including review and testing for prompts, skills and workflow definitions
- Cloud production experience, for example Azure or AWS, and container platforms such as Kubernetes
- Experience in a regulated or compliance-driven setting (SOC 2, ISO 27001)
Compensation and benefits
Pay: $90,000 – $250,000 yearly base salary.
- Competitive base salary, based on your role, experience, and location
- Remote-first and async by default, with shared core hours of 09:00–13:00 UTC
- A company laptop and the equipment you need to do your job
- Approved conferences covered: registration, travel, and accommodation
How we hire
Interviews with our leadership and engineering team. We ask candidates to walk us through real work they shipped.