- Design and optimize prompts for conversational AI personas and personalized user experiences
- Build and maintain RAG pipelines using vector databases such as pgvector
- Implement embedding-based personalization and content matching
- Work with LLM model routing based on quality, sensitivity, and cost
- Integrate Voice AI services such as TTS/STT APIs
- Implement AI safety, content guardrails, and crisis-language detection
- Build evaluation frameworks with golden sets and adversarial test cases
- Monitor AI output quality and production performance
- Collaborate with backend and mobile teams on latency, reliability,
and cost optimization
? ???????? ??????
✅ Solid Python
✅ FastAPI or similar async frameworks
✅ Production experience with LLM APIs – Claude, OpenAI, or similar
✅ Hands-on RAG & Vector Databases
✅ Prompt Engineering with evaluation-driven approaches
✅ LangChain / LangGraph or similar orchestration frameworks
✅ AI safety & content moderation
✅ LLM cost optimization – caching, model routing, token budgeting
✅ Understanding of failure handling, timeouts & vendor fallbacks