Kamyar Taeb

Built an automated student email-enquiry system to replace a manual workflow with faster, more consistent responses. Developed a production RAG pipeline improving retrieval quality and scalability. Shipped agentic LLM workflows automating multi-step tasks end-to-end, streamlining development with Python, Git, FastAPI, and Claude Code.

Developed an agentic career upskilling platform using service orchestration to deliver personalized learning and career recommendations. Built and orchestrated agentic services with tool calling, prompt engineering, and guardrails to enable reliable, safe, production-ready AI workflows. Developed a real-time conversational AI avatar with low-latency streaming and frontend integration. Designed evaluation pipelines to assess and improve the reliability and quality of agentic systems.

Trained a Persian-language model for automating inbound call handling, cutting call-centre staffing by 90% (about 1,000 to about 100 agents). Built end-to-end Python pipelines for preprocessing, training, and evaluation; delivered personalized AI for customer profiling and recommendation.