What will you do?
TL;DR: Use Python, embeddings, and LangChain to build an AI agent that keeps our Knowledge Graph truthful — solving one of the biggest challenges in enterprise AI: semantic drift.

The Mission - Generic AI models struggle to understand the specific DNA of a business. At Metagem, we solve this by distilling company knowledge (unstructured docs, database schemas…) into a structured Ontology and Knowledge Graph. This "Unified Semantic Layer" acts as the brain of the company. It allows us to deploy AI agents that don't just chat, but truly understand business logic and adapt to the specificities of every client.

Your Challenge - The reality of company data is messy: documents conflict and terminology drifts. To maintain a reliable Ontology, we need to detect when a document diverges from the established truth. You will build a tool to quantify this Semantic Alignment, working directly with the AI Lead and Tech Lead.

Depending on your background and interests, you will contribute across one or more of the following areas:

• Research & Evaluation
  Explore and assess different similarity metrics (e.g. evaluating whether cosine similarity is sufficient or if more advanced approaches are required). You will benchmark various embedding models, design experiments, and build test datasets that capture known inconsistencies to support robust validation.

• Development & Prototyping
  Design and implement a measurement agent using Python and LangChain. This includes crafting Cypher queries to retrieve relevant context from a graph database, implementing verification and feedback loops (for example, detecting and responding to similarity scores below a defined threshold), and running the agent against curated test datasets.

• Deployment & Integration
  Package the solution using containerisation, deploy it in a validation environment, and expose a FastAPI interface to enable interaction with the agent and support integration into existing workflows.

Who's team Metagem?

Founded in 2025, Metagem is a team of three founders building the infrastructure for the next generation of AI agents. Each founder brings deep expertise — AI engineering, technical architecture, and business development — and we’ve already worked as a tight project team on customer solutions, platform builds, sales demos, and even securing funding.

We’re also used to mentoring interns and junior engineers.
By joining us, you’ll work directly with experienced founders, see how a startup grows from the inside, and contribute to real product decisions from day one.

Who we're looking for?

Above all, we are looking for someone eager to learn, discover new technologies, and ship high-impact code. Concretely:

• Student in Computer Science or Engineering (Master’s level preferred)

• Solid Python and standard software engineering practices (Git flow, clean code)

• Understanding of Embeddings and RAG (Retrieval-Augmented Generation)

• Bonus: Experience with LangChain or graph databases

The Deal

• You’ll leave with a portfolio-ready AI agent project

• You’ll learn how to build production-grade AI components

• Internship (leading to potential full-time hire)

• Duration: 3 to 6 months (flexible)

• Start: From Feb. 2026 (flexible)

• Location: Wintercircus Ghent, Belgium