Agentic products & LLM architectures

Overview

I lead design and delivery of AI-native systems: autonomous agents, tool use, RAG, multi-provider LLMs, and the backends that keep them reliable in production.

Problem

Teams needed agents and LLM features wired into real tools, data, and workflows — not chat demos — with ownership that can ship and operate them.

Approach

• Define agent boundaries: tools, memory, evals, failure modes, and human-in-the-loop paths.
• Build multi-provider LLM layers across OpenAI, Anthropic, Gemini, and Groq.
• Implement RAG pipelines and orchestration on real APIs and data stores.
• Pair AI surfaces with NestJS and Go services, auth, and cloud deployment.
• Lead architecture reviews, delivery, and mentoring across a team of five or more.

Outcome

• Made agentic and LLM delivery a first-class product capability at the studio.
• Shipped production-minded systems with modular APIs and maintainable code.
• Shared patterns for CQRS backends, agent workflows, and cloud operations.

Stack