Meegrow Labs – AI Engineering Internship (Project Horizon)
8–12 week, high-intensity AI engineering internship working on a real Project Horizon research initiative: turning natural language building descriptions into zoning-compliant 3D massings.
About the Project
An international R&D effort to build production-grade AI systems for architectural design automation.
What We're Building
The Project Horizon system is an end-to-end AI pipeline (LOD 100) that converts natural-language architectural prompts into fully structured 3D building massings.
The system converts:
- Natural language prompts → Structured architectural briefs
- Briefs → Zoning-compliant building boundary boxes
- Boundaries → LOD-100 architectural massings (3D envelopes with floor divisions)
Project Deliverables
- A 2,000-sample dataset with prompts, briefs, zoning envelopes, and 3D massings
- Three AI models: NLP compiler, generative 3D model, and QA critic
- Automation scripts, evaluation frameworks, and reproducible environments
- Research-grade documentation and experiment tracking
Team & Collaboration
Interns work directly with a postdoctoral researcher and a senior AI architect based in Germany. This is real R&D work with international collaboration standards.
Role Overview
Position Details
- Title: AI Engineering Intern (Project Horizon)
- Positions: 4 interns
- Target: Strong 3rd/4th-year CS students or recent CS graduates with relevant skills
- Location: Remote or Gurgaon (hybrid)
Technical Environment
- NVIDIA GPU servers (Blackwell-ready)
- Python, PyTorch, NumPy, pandas
- Git/GitHub workflows with CI
- Ubuntu 24.04 LTS
Duration & Timeline
8–12 weeks | Approximately 14 December 2025 – 15 March 2026
Flexible start date depending on team setup. Expect 30-40 hours per week of focused work.
What You'll Work On
01 / Dataset Engineering & Automation
- Build Python scripts to assemble and validate a 2,000-sample dataset
- Implement schema validation, cleaning pipelines, and QA checks
- Work with geometric attributes and site metadata from external sources
- Create reproducible data preprocessing workflows
02 / Model Pipelines
- Preprocessing for the Prompt → Brief → Boundaries model (normalization, tokenization, structured JSON)
- Assist with training and evaluating NLP models using PyTorch
- Prepare and process 3D data formats (voxel / SDF / point clouds) under guidance
- Run GPU training jobs, monitor logs, and manage checkpoints
03 / QA & Evaluation
- Implement rule-based geometric checks (envelope violations, area mismatches)
- Build critic/evaluator components for comparing 3D massing quality
- Create automated testing frameworks for model outputs
04 / Engineering Infrastructure
- Write modular, tested Python code with proper documentation
- Work with GitHub branches, PRs, and CI workflows
- Package environments (conda / Docker) and support experiment tracking
- Maintain reproducibility standards for research code
You'll Be a Good Fit If...
Core Requirements
- Strong Python skills (modular code, data handling)
- Comfortable on Linux and the command line
- Familiarity with deep learning fundamentals and training loops
- Experience with PyTorch or similar frameworks
- Basic Git/GitHub experience
Nice to Have (Bonus)
- Experience with 3D data (OBJ, GLB, voxels, point clouds)
- Exposure to computational geometry or BIM concepts
- Experience with diffusion models or transformers
- Blender scripting or 3D programming
- Docker, FastAPI, or experiment-tracking tools
What You'll Learn
- End-to-end AI system design: dataset → model → evaluation → deployment
- Working with large NVIDIA GPU servers and optimizing training workflows
- 3D generative AI and geometric deep learning techniques
- How real AEC (Architecture, Engineering, Construction) workflows influence model design
- Research-grade code quality: CI/CD, reproducibility, and proper documentation
- International collaboration and R&D best practices
Timeline & Process
Applications Open
Submit your application through the form
Applications Close
Final deadline for submissions
Shortlisting & Interviews
Technical assessment + interviews with selected candidates
Internship Starts
8-12 weeks of intensive R&D work (until mid-March 2026)
How to Apply
Application Process
- Click "Apply Now" and fill in the application form
- Provide: Basic details, LinkedIn/GitHub, short answers, and upload your CV
- We'll email shortlisted candidates with a brief technical task + interview details
Frequently Asked Questions
Yes, this is a paid position. Compensation details will be discussed during the interview process based on experience and commitment level.
Yes, the internship can be done fully remote. However, if you're in or near Gurgaon, we offer a hybrid setup with occasional in-person collaboration opportunities.
This is a high-intensity internship expecting 30-40 hours per week of focused work. You'll be contributing to real R&D, so consistent availability and meeting deadlines is critical.
No, prior 3D experience is not required but is a strong bonus. We'll provide guidance and resources. What matters most is strong Python skills, willingness to learn, and comfort with technical depth.
Yes, upon successful completion, you'll receive a letter of recommendation and internship completion certificate. More importantly, you'll have real R&D contributions and potentially co-authorship opportunities.
Absolutely. High-performing interns who demonstrate strong engineering skills and research mindset will be considered for extended roles or full-time positions at Meegrow Labs.
Ready to build real 3D AI,
not toy notebooks?
This isn't a "certificate" internship. If you're serious about AI engineering, want to work on cutting-edge research, and are ready to ship production-grade code—apply now.
Apply NowApplications close on 30 November