Kyrylo Khazanzheiev
I’m Kyrylo Khazanzheiev, an AI Engineer focused on building secure LLM applications, RAG pipelines, and agentic workflows that integrate cleanly with real business systems.
About
I’m Kyrylo Khazanzheiev, an AI Engineer focused on building secure LLM applications, RAG pipelines, and agentic workflows that integrate cleanly with real business systems. Over the past four+ years, I’ve owned end-to-end delivery—from architecture and Python/FastAPI services to evaluation, monitoring, and production-readiness—so what ships is both reliable and supportable.
I enjoy turning complex requirements into practical user journeys, including document assistants and bounded tool-calling agents with human-in-the-loop review. I’ve worked across Azure and AWS (with Terraform, least-privilege controls, and automated tests), and I’ve integrated platforms like ServiceNow, Jira, and GitHub to deliver measurable outcomes in real operational environments.
- 5 years' experience
- Joined October 2026
Industry experience
- Software & Internet
- Professional Services
- Computers & Electronics
Work experience· 6 roles
Designed and delivered a governed case-intelligence workflow on Azure Databricks using synthetic data with Azure OpenAI and Azure AI Search. Implemented structured extraction and grounded RAG with a bounded Case Officer agent. Provisioned Azure and Databricks resources with Terraform and Microsoft Entra RBAC/workload-identity federation, removing secrets from infrastructure state. Added validation gates, data-quality reconciliation, evidence-based refusal when information was insufficient, and documented security gaps, cost drivers, and pre-production controls.
Developed an agentic document-processing workflow across Python, FastAPI, AWS, and Terraform with validation, repair stages, processing-status visibility, and human-in-the-loop review. Supported separate development and production environments, CI/CD promotion, automated deployment, and operational retry controls. Achieved 95.3% reconstruction quality while reducing model cost by ~61% and end-to-end processing time by 28% versus an internal all-GPT-5.5 baseline.
Built an AI-agent system integrating ServiceNow, Jira, and GitHub through MCP using Python/FastAPI services and controlled tool calling. Implemented incident listing, retrieval, creation, updating, search, and work-note workflows with write operations disabled by default and enabled only via explicit configuration. Added Pytest and Playwright end-to-end coverage for UI, API, and MCP services, with multi-service CI and troubleshooting documentation.
Education
Hire Kyrylo Khazanzheiev today
Post a job and invite Kyrylo, or contact them directly. No hidden commission.