Alisher Zhanayev

Builds production RAG assistants over internal knowledge bases for support and operations teams, and an agentic answering flow for leadership questions. Implements guardrails for a regulated environment including access scoping, PII filtering, and audit trails.

Built and maintained backend services for core platform flows, focusing on reliability and clean integration boundaries. Integrated third-party services and automated internal operational workflows, cutting manual handling by ~30%.

Designed and shipped the company's first production LLM assistant with an orchestrator routing messages to FAQ RAG, recommender, and open conversation skills. Built the recommender end to end, added long-term user memory, and a secondary LLM verification layer cutting hallucinated answers by ~60%. Improved retrieval quality and reduced p95 latency from ~4.2s to ~1.8s. Led the backend team, growing

Improved delivery ETA accuracy from 84% to 91% by optimizing routing algorithms and backend logic. Decomposed a monolith into 8 Go-based microservices, raised test coverage from 90% to 95%, and reduced failed payment rate from 6% to 4%.

Reduced report generation time from ~1 hour to ~15 minutes by optimizing database queries and processing pipelines. Cut order delivery time by improving route-planning logic and automated internal workflows.

Completed the Hugging Face AI Agents Course in 2026.

Bachelor's degree from the Faculty of Information Technologies at Al-Farabi Kazakh National University, completed from 2017 to 2021.