MCP & Memory Intern
About the role
This internship focuses on making AI agents more useful by giving them the right tools, context, and memory. You will work on the Model Context Protocol (MCP) and agent memory to help agents discover tools, call them safely, and track what they have learned. The goal is to build practical, clean implementations that others can use. We value curiosity and the ability to figure things out over a perfect resume.
What you'll do
- Build and improve MCP servers and clients, including tool definitions and error handling.
- Experiment with agent memory, such as how data is saved and retrieved.
- Wire tools and memory into an agent loop to identify where it gets slow or confused.
- Make small, clear changes like documentation, tests, and fixes to remove friction for the team.
- Turn rough ideas into working code and tighten them based on results.
- Pair with team members to learn the system and eventually own your own tasks.
What you'll need
- Comfort reading and writing Python.
- Curiosity about how AI agents work, including prompts, tools, and context windows.
- Interest in MCP protocols and agent memory, such as storage and state management.
- Ability to take a vague problem, try a small version, and share what you learned.
- Clear communication skills and a willingness to ask for help when stuck.
- Good engineering habits like writing readable code and small tests.
- Ability to take feedback and adjust your approach.
- Strong plus: Experience building or using an MCP server or client.
- Strong plus: Experience with agent frameworks, tool calling, or conversation state.
Location & details
- Location: Boston, Massachusetts (Remote).
- Employment: Full time, paid internship.
- Term: Rolling.