# Machine Learning Engineer > Raydar · San Francisco, United States · Full-time · Posted 2026-11-07 **Salary:** USD 160,000–210,000 **Workplace:** on_site ## Description About the company Our client is a property technology company. The role Raydar is recruiting for this role on behalf of our client. Improve the accuracy of AI agents that analyze construction documents against regulatory requirements, working through model training, experiment design and evaluation. The role suits an engineer who can move quickly, test ideas rigorously and extend the product into new building types. What you'll do \- Spend most of your time raising the accuracy of multimodal AI agents, including model post-training and automated research. \- Design experiments, build them out and analyze results on internal evaluation frameworks. \- Handle a range of supporting work such as light data engineering, tooling for the engineering team, and web application improvements. \- Develop new agents that extend the current product to commercial and multifamily housing nationwide. \- Speak with users on a weekly basis. ## Requirements What we're looking for \- Works independently, moves at high speed and holds evaluation work to a rigorous standard. \- Either a strong software engineer transitioning into AI research, or an AI or machine learning engineer already working in the field. \- At least 3 years of relevant experience. \- Shared interest in housing supply and land-use reform. \- Willingness to work long weeks, roughly 55 hours, including some weekends. ## Benefits Compensation and benefits \- Base salary: USD 160,000 to 210,000 per year \- 401(k) retirement plan with pre-tax and Roth contributions \- Health insurance, fully paid \- Employer health savings account contributions \- Commuter benefits \- Wellness benefit \- Meals provided at the office Location and work model \- San Francisco, California, United States \- On-site \- Full-time ## Apply [Apply at Raydar](https://apply.workable.com/raydar/j/DB5C68DD13/apply) --- Powered by [Workable](https://www.workable.com)