AI is the core of what Kive does. You'll be the second AI engineer, working directly on the pipelines and models that power image and video generation agents for thousands of customers. This is production AI engineering — reliability, cost, speed, and quality all matter equally.
You will
- Improve and expand our generation capabilities — quality, speed, and cost per generation
- Integrate and evaluate new models fast when they drop — and manage vendor dependencies so upstream changes don't break the product
- Build evaluation pipelines to measure quality objectively, not just by vibes
- Optimize inference infrastructure and GPU costs (this is existential at our ARPA)
- Work closely with product to turn AI capabilities into features customers love
You
- Have shipped model-powered features to real users in production — not just trained models.
- Have strong ML engineering fundamentals and have worked with generative models in production.
- Care about cost, latency, and reliability as much as output quality.
- Stay close to the frontier and move fast when new things drop.
- Want to own this end-to-end at a small company, not be a cog at a big one.
In your first
Week
- Get hands-on with the full generation pipeline from prompt to output.
- Run the product, generate content, and form your own view on where quality and speed fall short.
- Understand our inference stack, costs, and vendor dependencies.
3 months
- Meaningfully improve an agent’s output quality, speed, or cost — and have a clear plan for the others.