Research Intern, Special Projects (AI)

TL;DR

Research Intern, Special Projects (AI): Build data pipelines, design vision-language models, and implement agentic systems for AI-assisted home design editing from natural-language instructions with an accent on structured output generation, layout validity, and edit correctness. Focus on training/fine-tuning LLMs, developing evaluation suites, running ablations, and prototyping product integrations.

Remote (United States)

Company

VC-backed startup changing how new homes are designed and built with support from top venture capital and strategic investors.

What you will do

• Build data pipelines to extract and curate structured datasets from existing design data
• Design and implement models for vision-language understanding, structured output, and agentic editing of home designs
• Execute training, fine-tuning, and post-training pipelines including preference optimization and RLHF-style methods
• Develop evaluation suites and error taxonomies for layout validity, instruction-following, and edit correctness
• Run ablations, scaling experiments, and maintain research reproducibility
• Prototype end-to-end AI-assisted design workflows and document findings via reports, demos, and publication drafts

Requirements

• Active enrollment in a PhD program in Computer Science, Machine Learning, or related field at a U.S. institution
• Strong Python skills and hands-on experience with deep learning training workflows (PyTorch preferred)
• Solid understanding of at least two of: computer vision, transformers, LLMs, representation learning, ML experimentation
• Demonstrated research ability through publications, preprints, open-source code, or equivalent
• Comfort with research/engineering stacks like Roboflow, Modal, W&B

Nice to have

• Experience with vision-language models, multimodal foundation models, or grounded document understanding
• Building or fine-tuning agentic LLM systems for structured state reasoning and edits
• Familiarity with post-training techniques: RLHF, DPO, preference optimization
• Multi-GPU training and large-scale experimentation
• AEC data/workflows knowledge including CAD/BIM, floor-plans

Culture & Benefits

• Remote-first since 2018 with flexible hours
• Competitive salaries, significant equity
• Comprehensive medical, dental, vision coverage, unlimited PTO, maternity/paternity leave for full-time U.S. employees
• Short/long-term disability plans, 401K
• Home office stipend