This may include multi-stage workflows such as plan → build → review, model integrations, structured outputs, tool calling, streaming, caching, prompt improvements, and new AI agents.

You’ll also contribute to production features, bug fixes, and pipeline development in the main codebase.

🧪 Evals & Quality

Help us understand whether each product change actually makes the AI better.

You’ll create practical evaluation systems for generated applications, business plans, marketing content, AI reliability, and release regressions. Depending on the task, you may use automated checks, browser tests, LLM judges, human-reviewed datasets, and side-by-side comparisons. When something fails, you’ll help find the cause, improve the prompt or pipeline, ship the fix, and add a regression test so the same issue does not return.

The goal is simple: catch problems before users do and make product decisions based on evidence.

🏆 What Success Looks Like

• New AI features and improvements reach production regularly
• AI quality becomes visible and measurable
• Regressions are caught before release
• Generated products become more functional and reliable
• Evaluation findings consistently turn into shipped fixes
• Model quality, cost, and latency stay under control

Over time, you’ll become a key technical partner for the founders, product team, and engineers shaping Ventora’s AI platform.

🤝 What You Bring

Required:

• 3+ years of experience in AI/ML, applied data science, or backend engineering
• Hands-on experience building production applications with LLMs
• Experience with model APIs such as OpenAI, Anthropic, or OpenRouter
• Understanding of prompts, agents, tool calling, structured outputs, and multi-stage AI workflows
• Ability to turn unclear questions like “Is this version better?” into practical, measurable tests
• Strong ownership, product thinking, and clear communication
• Regular use of AI coding tools such as Codex, Claude Code, Cursor, or similar
• Comfortable working in Linux environments and with Docker

You don’t need experience with every framework or evaluation platform. We care more about strong engineering fundamentals, curiosity, and the ability to learn.