1. Build and iterate on LLM-powered features across the studio's portfolio, prompt pipelines, and agent workflows, RAG systems, structured output, evals, and full-stack integrations. 2. Write, test, and refine prompts; evaluate outputs against rubrics; reduce hallucinations and improve grounding. 3. Integrate Claude/OpenAI/open-source models via SDKs and APIs; wire up tool use, function calling, and retrieval. 4. Contribute to our autonomous multi-agent QA engine, add detectors, triage findings, and close the feedback loop from bugs back into agent learning. 5. Own small-to-medium features end-to-end: scope, build, ship, monitor. 6. Write clear technical notes and demo videos for non-engineering stakeholders.