FortyGuard · Case study
AI agents and MCP tools for temperature intelligence.
FortyGuard helps users understand environmental conditions and heat exposure through temperature mapping and analysis. I work on AI assistants that connect questions about this data to backend tools and the dashboard.
- Role
- Software Development Engineer Intern
- Context
- July 2026–Present
My contribution
I develop Python/FastAPI agent backends using LangGraph for agent workflows, working on state and context, tool calling, backend/API integrations, and responses grounded in tool results. My work also includes streaming responses, frontend integration, knowledge-base improvements, reliability debugging, and end-to-end testing.
Connecting models to tools
I use FastMCP for server and tool integrations through Model Context Protocol (MCP), which lets AI applications access backend tools. These integrations connect LangGraph agent workflows to internal services. The engineering problem extends beyond generating an answer: the agent needs the right context, a clear tool contract, and controlled actions based on the returned results.
Temperature-aware routing
The ML team developed the temperature-aware routing algorithm. I work on connecting it to backend services and map interfaces through service integration, contracts, task execution, hardening, testing, and frontend route rendering.
Reliability and validation
Reliability debugging and end-to-end testing are part of my work across the agent backend and frontend. The integration boundaries—model, tool, service, and dashboard—are where I focus on keeping execution and responses consistent.
Ongoing work
This is ongoing, collaborative work across the agent backend, internal services, and dashboard. Internal implementation details and company material remain private.
Technologies & references
- Python
- FastAPI
- LangGraph
- FastMCP
- Model Context Protocol
- Streaming
- Frontend & map integration