Skip to content
← All casesAgent InfrastructureAI

AI Agent Platform

The infrastructure layer under agentic systems: LangGraph (stateful graphs) + LiteLLM (one gateway to providers) + A2A protocol (agent-to-agent interop)

At a glance
LangGraph + LiteLLM + A2A protocol

Founder, architect & solo developer

solo · solo

  • Python
  • LangGraph
  • LiteLLM
  • A2A
  • FastAPI

Expand the deep-dive ↓

Problem

What doesn't work

Building agentic systems from scratch each time is expensive: your own state management, your own calls to each provider, your own inter-agent protocol. No reusable layer.

Solution

Architectural approach

A substrate platform: LangGraph provides stateful graphs for multi-step agents; LiteLLM unifies calls to Claude/GPT/DeepSeek/Gemini behind one interface; the A2A protocol standardizes inter-agent interaction. On top — a gateway with routing, tracing and cost control.

My role & contribution

Founder, architect & solo developer

Designed and built a platform to compose and run LLM agents: LangGraph for stateful multi-step graphs, LiteLLM as a unified gateway across providers, Google's A2A protocol for agent-to-agent interop. Gateway + UI, per-agent routing, tracing and cost control.

Ready to discuss?

If you need an architect who builds autonomous AI systems — reach out.

Serbia-based · CET/CEST timezone · EU-aligned working hours · International contracts experience