Internship
Artificial Intelligence Intern @ Geomatys
Developed a geospatial LLM chatbot for EpiWise, an epidemic-risk application, enabling non-scientific users to interact with map layers, dates, model parameters, screenshots, and generated reports through an agentic interface.
Highlights
- Contributed to the EpiWise project context during the period in which the project won the ESA Phi-Lab Grand Marathon first prize according to the CV.
What I did
- Implemented a ReAct agent with LangChain and LangGraph to interact with the Examind geospatial API and update map visualizations.
- Built LLM tools for WFS and WMS APIs to retrieve geospatial context and change the displayed map state.
- Integrated chatbot actions for changing layers, dates, and model parameters in the geospatial interface.
- Implemented screenshot and report-generation workflows so users could receive a final report with the conversation history, map screenshots, and results.
- Started implementation work around LLM protocols such as MCP and A2A for future interoperability.
How it unfolded
AI internship at Geomatys centered on a geospatial LLM chatbot for EpiWise.
-
Mar 2025 Onboarding
AI internship started
Started the Artificial Intelligence internship at Geomatys in Montpellier.
-
Apr 2025 ~ Implementation
LangGraph/LangChain ReAct agent
Implemented a ReAct agent to interact with the geospatial application and Examind APIs.
-
May 2025 ~ Implementation
Geospatial tool integration
Added WFS/WMS tool calls and richer map context for chatbot answers.
-
Jun 2025 ~ Delivery
Map screenshot and report workflow
Enabled the chatbot to manipulate map state, take screenshots, and generate user-facing reports.
-
Aug 2025 Delivery
Internship completed
Completed the 6-month AI internship and delivered the chatbot work.
Stack
- Python
- LangChain
- LangGraph
- ReAct agents
- Mistral
- Examind
- WFS
- WMS
- MCP
- A2A
- Geospatial APIs