What are the responsibilities and job description for the Founding ML/AI Engineer position at Arcway?
About Arcway
Arcway is an architectural design tool where buyers can design and walk through a photoreal home instantly in their browser. Homebuilders use Arcway to close faster and sell more homes. We’re a small team of ex-architects and product builders from Figma, Coinbase, Xsolla and Modern Health. We’re backed by Indie, Ritual, Soleio, Jorn Koen, and other top angels.
Role & Responsibilities
Arcway ingests unstructured design files, and turns them into production-ready 3D homes. Today this process is slow, error-prone, and requires architectural intuition to resolve contradictions, missing details, and layout issues. You will design ML systems that eliminate these bottlenecks.
You will also help automate home staging (e.g. faucets, lighting fixtures): matching SKU data we receive from a customer to create performant, game-ready 3D assets from our internal library, manufacturer datasets, or generative pipelines.
- Convert unstructured design files into accurate 3D geometry data.
- Turn product references into usable 3D assets.
- Optimize 3D assets using ML/GenAI.
- Build end-to-end ML pipelines that ingest, parse, extract, validate, and output structured JSON.
- Create feedback loops where human corrections feed into a retraining cycle.
Tools & Languages
Core ML / AI Tooling
- Python
- PyTorch (or TensorFlow/JAX) for training and running models
- Transformer models for multimodal document understanding, for example LayoutLMv3
- Vision models for recognizing floor plans, symbols, and annotations, for example Detectron2
- Diffusion or generative models for simple shape completion or reconstruction
Geometry, CAD, and Floor-Plan Understanding
- CAD parsing experience using libraries such as ifcopenshell
- Knowledge of geometric primitives, 2D to 3D inference, and computational geometry using tools such as Open3D
- Experience extracting symbols and room features from PDFs or images with OpenCV
Document Parsing and Structured Data Extraction
- OCR and layout extraction using tools such as Tesseract
- Embedding and similarity search systems such as FAISS for asset lookup
We're Looking For
- People that want to work closely with co-founders at a fast-paced startup
- Strong communication skills, we expect you to be able to deeply describe a technical problem you worked on in layman's terms
- Deeply care about data quality and consistency
- Comfortable working with highly variable builder documents that contain contradictions or missing information
Bonus (not required)
- Interest in 3D generative AI, for example Meshy or TripoSR
- Experience generating synthetic data for indoor layouts
- Personal interest in architecture, BIM, or CAD systems
What We Offer
- Be an early hire building the backbone of a fast-growing AI company.
- Influence how data, AI and 3D design intersect in the built world.
- Awesome benefits: https://golden-magnolia-e09.notion.site/Arcway-Employee-Benefits-27ab7624659980daa022febd0c4cb495?pvs=74.
- Competitive salary, meaningful equity, and hybrid flexibility.