Senior AI Engineer

The Innovation team works across the whole organization, engineering and beyond, as part of ZOOP's journey to becoming AI-native. We build internal tools first and adopt them ourselves, so the feedback loop is short: if something is annoying, we feel it the same day. Current projects cover the model router, the desktop coding client, the developer installer and CLI, and Slack and Linear automations. We keep a running log of experiments and their results, and we prefer small, reversible steps over big launches.

Responsibilities:

• Maintain and improve the team's projects: the internal model router (Go), the desktop coding client and its CLI (TypeScript, Node, Electron), the developer installer and update service, and the repo-aware Slack and Linear bot.
• Run the router day to day: add models and providers, keep authentication and quotas working, attribute usage and cost per team, and maintain the numbers that show whether routing decisions actually pay off.
• Evaluate new models, coding agents, and developer tools continuously. Write short, concrete notes on what works, what it costs, and what we should adopt or skip.
• Build automations that other teams feel: Slack and Linear bots, GitHub apps, webhooks, and small services that remove repeated work.
• Work with non-engineering teams, for example, operations, support, sales, finance, and people teams, to find repetitive work and build AI-native workflows for it.
• Help teams across the organization adopt AI-native ways of working: onboard people, collect friction, and turn it into fixes, documentation, and features.
• Drive ZOOP's AI-native journey with numbers: define the metrics that show how AI is used across the organization (adoption, time saved, cost, quality), track them, share them, and improve them over time.
• Run regular hands-on sessions that upskill teams across the company: how to use AI tools well, safely, and within budget, and how to spot work worth automating.
• Own the unglamorous parts: secrets handling, budgets, dependencies, releases, and uptime.
• Grow the team: help interview, onboard, and mentor. The plan is one or two hires in your first year, and you will have a real say in who and how.

First 30 days:

• Your own machine is fully on the toolkit, and you are doing your daily work through it.
• You have shipped your first small fixes across the team's projects.
• You understand the router's role, its costs, and who depends on it.
• You have written your first experiment note: what you tried and what you found.
• You have met a few non-engineering teams and know where their most repetitive work is.

By 60 days:

• You own one project end-to-end, such as a router capability or a stage of the Slack and Linear bot.
• At least one automation you built is in use by people outside the team, ideally including a non-engineering team.
• New models and tools are being evaluated on a schedule rather than by accident, with results the team can act on.
• A first version of the organization's AI-native metrics is tracked and shared, and you have run your first upskilling session.

By 90 days:

• You are coordinating with two or three teams on adoption, with at least one outside engineering, and adoption is measurable: usage, cost, or hours saved.
• You are reviewing others' code and designs, and the experiments log has a rhythm you own.
• You are helping interview for the team's first hires.
• Upskilling sessions run on a regular cadence, with a simple way to see whether attendees use the tools afterwards.

At 1 year:

• You lead a small team of two or three, and the hiring you drove has landed well.
• AI-native workflows are in daily use across engineering and in several non-engineering teams, with real numbers to show for it: hours saved, spent controlled, and agents trusted with real work.
• There is a documented evaluation loop for new models and tools, and other teams come to you before picking one.
• The innovation pipeline has a cadence: try, measure, adopt or drop, and tell everyone what happened.
• The organization's AI-native metrics have improved over the year, you report them regularly, and teams across the company have been upskilled through your sessions.

Requirements:

• 5 to 8 years of professional software engineering experience.
• You use agentic coding tools such as Claude Code, Codex, or Cursor daily, and you have formed opinions about how to get real work done with them.
• You have built applications on LLM APIs and know the practical side: model routing, evaluation, prompt and tool design, and cost control.
• Strong TypeScript/Node and Go.
• You have built automations and integrations: Slack apps, Linear or similar trackers, GitHub apps, webhooks, and internal workflow automation.

Nice-to-haves:

• None of these are required. If you have them, tell us.
• You have operated an LLM gateway or proxy in production: auth, quotas, per-team accounting, and failover between providers.
• You have shipped or maintained an Electron desktop app or a developer CLI that engineers install themselves.
• You have run structured evaluations of models or agents, not just informal spot checks.
• You have run workshops or training sessions and enjoy teaching people outside engineering.
• You have automated workflows for non-engineering teams, or tracked adoption metrics for a tool rollout.

Skills

CLI, Electron, GitHub, Go, LLM, LLM API, Node.js, Slack, TypeScript, Webhooks