About Hedral
Hedral is redefining building design through automated, multi-disciplinary engineering.
We develop and operate a proprietary platform that accelerates design schedules, enabling us to deliver the output of traditional AEC firms ten times our size. We are fully licensed to stamp designs as Engineer of Record and take responsibility for highly-consequential projects, from mission-critical federal facilities to complex commercial infrastructure.
We are scaling quickly with backing from leading venture capital including Khosla Ventures, Valor Equity Partners, and Tishman Speyer.
The Role
We are seeking a Machine Learning Scientist to join our core technical team and help build the next generation of AI-driven design systems.
This role sits at the intersection of machine learning, spatial reasoning, and engineering, and involves designing, implementing, and deploying models that operate directly on complex architectural and engineering data. You will work on problems spanning computer vision, language modeling and agentic workflows, graph-based modeling, and reinforcement learning, with direct impact on real-world, high-stakes projects.
This is a high-ownership role where you will contribute both to research direction and production systems, helping shape the technical foundation of Hedral’s platform.
Location
Austin, TX or New York, NY - Primary preference for Austin or NYC; remote-eligible for exceptional candidates with a proven track record.
Core Responsibilities
Design and develop machine learning models for understanding and reasoning over architectural plans, engineering documents, and spatial data.
Build and deploy systems using vision, language, and graph-based models to process 2D drawings, 3D geometries, and structured engineering data.
Develop and train surrogate models for structural and MEP simulations to accelerate design evaluation.
Design and implement reinforcement learning and optimization systems for automated and combinatorial design problems.
Build robust data pipelines and training infrastructure for large-scale, domain-specific datasets.
Collaborate closely with engineering and product teams to integrate ML models into production workflows.
Lead experimentation, evaluate model performance on real-world use cases, and iterate rapidly to improve system quality and reliability.
Contribute to the overall system design, including modeling decisions, architecture choices, and scaling strategies.
Requirements
Master's degree in Machine Learning, Computer Science, Engineering, Applied Mathematics, or a related field; PhD is preferred but not required.
3+ years of industry or equivalent research experience building and deploying ML systems.
Proficiency in Python and deep learning frameworks such as PyTorch (preferred), JAX, and TensorFlow.
Strong understanding of modern ML techniques, including deep learning, reinforcement learning, and model evaluation.
Experience designing and implementing end-to-end ML systems, from data processing to model deployment.
Strong publication record with the ability to build on recent ML literature, ship production-ready systems, and operate independently in ambiguous environments.
Bonus Qualifications
-
Experience with one or more of:
Computer vision (e.g., segmentation, detection, plan understanding)
Language models and multimodal systems
Graph neural networks or structured data modeling
Reinforcement learning or optimization
3D or geometric deep learning
Experience building surrogate models or working with physical simulations.
Familiarity with large-scale data processing, distributed training, and ML infrastructure.
Background in Engineering, Architecture, or AEC with hands-on experience processing complex engineering data or spatial representations (CAD/BIM), and familiarity with relevant software (e.g., SAP2000, ETABS, Revit), reinforced concrete/steel design, and building codes (e.g., ASCE 7, ACI 318, AISC 360).
Experience with MLOps practices, including experiment tracking, model deployment, and monitoring.
How we operate
Ownership & Intensity: We operate with high autonomy and a shared sense of urgency to redefine an entire industry.
Bias for Action and Efficiency: We prioritize automation over manual effort to solve high-consequence problems, shipping quickly and iterating continuously.
Data-Driven Decisions: Hedral’s core advantage is not just automation, but that our system learns from real, stamped engineering work. Every project feeds back into the model.
What we offer
Competitive Compensation and Benefits.
Environment for growth.
Hybrid and flexible work environment.
Opportunity to work on high-impact, real-world problems at the intersection of AI and engineering.
A fast-paced environment with significant technical ownership and growth.