Чем предстоит заниматься
Industrialize prototypes into production services on Azure — Azure AI Foundry / Azure OpenAI — from build-ready pack to a system real users depend on in weeks
Build agentic systems properly, choosing orchestration frameworks, RAG pipelines, vector DBs, knowledge graphs and MCP-based tool integration by need and engineer them for change
Build the eval harness first: golden sets, regression evals and guardrail tests wired into CI, with quality measured on every change
Engineer the guardrails, including input/output filtering, grounding and citation, PII protection, rate limits and human escalation paths
Deliver full stack services in Python and/or Java Spring Boot along with TypeScript/Angular front ends
Run production engineering end-to-end, covering CI/CD, observability with traces on every LLM call, cost and latency management and model-version churn absorbed by design
Build security and compliance in, respecting data classification boundaries in prompts, stores and logs, externalizing secrets and making every AI decision auditable
Iterate from real usage through hypercare, tuning and fixes based on evidence, and package patterns that worked for the next pod