Research Internship / Work-Study

JobsCloseBy Editorial Insights

NeoBIM GmbH is offering a remote internship/work-study in Germany for Master’s students to explore Generative Flow Networks applied to architectural design and spatial generation. The role involves implementing GFlowNet based generative models, framing architectural layouts as sequential decisions, crafting reward functions, generating multiple design options, and evaluating diversity, feasibility and performance, with opportunities to publish and contribute to a pipeline. Key requirements include Python, PyTorch, fundamentals of ML and RL, a passion for computational design, and the ability to work independently; familiarity with graph libraries (NetworkX, PyTorch Geometric), geometry tools (shapely, trimesh), and optional Rhino/Grasshopper is a plus. Apply by highlighting relevant projects, share a code sample or thesis idea, and contact Felix at +49 176 95422094.

neoBIM is a well-funded start-up software company revolutionizing the way architects design buildings with our innovative BIM (Building Information Modelling) software. As we continue to grow, we are building a small and talented team of developers to drive our software forward.

This internship focuses on the application of Generative Flow Networks (GFlowNets) to architectural design and spatial generation problems. The role involves developing generative AI models that can sample diverse, constraint-aware architectural solutions, supporting early-stage design exploration and research-driven prototyping.

The position is suitable for Master’s-level students interested in the intersection of architecture, AI, and computational design, and may be aligned with academic research or thesis work.

Tasks

Implement GFlowNet-based generative models for architectural design tasks

• Represent architectural layouts, massing, or spatial graphs as sequential decision processes

• Design and test reward functions encoding architectural constraints and objectives

• Generate and evaluate multiple architectural design alternatives

• Analyze results in terms of diversity, feasibility, and performance

• Compare GFlowNet approaches with baseline generative or reinforcement learning methods

• Contribute to documentation, experiments, and research deliverables

Requirements

Required Skills

• Experience with Python and PyTorch

• Familiarity with machine learning and reinforcement learning concepts

• Interest in computational architecture or generative design

• Ability to work independently on research-driven tasks

• Strong analytical and documentation skills

Technical Stack

Machine Learning

• Python

• PyTorch

• NumPy

Generative Modeling

• GFlowNet implementations

• Reinforcement learning fundamentals

Computational Design

• Graph libraries (NetworkX, PyTorch Geometric)

• Geometry processing tools (shapely, trimesh)

• Optional: Rhino / Grasshopper

Development

• Jupyter Notebooks

• Git / GitHub

Benefits

• Hands-on experience with state-of-the-art generative AI models

• Exposure to AI-driven architectural design workflows

• Development of a research-grade generative design pipeline

• Opportunity to contribute to academic or applied research outputs

• Potential alignment with a Master’s thesis or capstone project

If you are passionate about building cutting-edge software and want to be part of a company that is transforming the architecture industry, we would love to hear from you. +49 176 95422094 Felix