About the company
Our client is a fast-growing software company.
The role
Raydar is recruiting for this role on behalf of our client. Own the modeling side of a software platform, moving ideas from early experiments into dependable production use. The work spans adapting large language models, preparing data, and building ways to judge model quality, alongside researchers and systems engineers.
What you'll do
- Adapt large language models to specialized technical tasks using supervised learning and reinforcement-based methods.
- Create and maintain the data flows that feed training, including sourcing, cleanup and artificial data creation.
- Develop test suites and scoring methods that show how accurate and dependable the models are on realistic tasks.
- Plan and run experiments, study the outcomes and refine approaches in short cycles.
- Partner with researchers, systems engineers and domain specialists to ship new models into production.
- Track new academic and open-source developments and bring useful ideas into the team's models.
Requirements
What we're looking for
- Hands-on experience applying machine learning to hardware design data or electronic design automation, such as netlists, RTL, surrogate models or graph neural networks.
- Experience taking a model end to end: gathering data, post-training it, and assessing its accuracy and reliability.
- Prior machine learning work in electronic design automation, in academia or industry.
- Experience creating scoring harnesses and data pipelines that support large training runs.
- Practical background in customizing large language models after pretraining, for example with supervised or preference-based methods.
- Publications at leading EDA or machine learning conferences.
- A university degree in an engineering or computing discipline, ideally with a machine learning emphasis.
- Strong Python and PyTorch skills, plus familiarity with widely used LLM libraries.
- A self-directed working style and enthusiasm for ambitious, open-ended problems.
Bonus points
- Experience working with customers to collect requirements and turn technical solutions into products.
- Systems-level knowledge of inference, including C++, CUDA, GPU architecture, and model deployment and optimization.
- Experience with inference serving frameworks, distributed training or inference, and low-level debugging and profiling.
- Exposure to agent-based workflows, code-generating models or hardware description languages.
Benefits
Compensation and benefits
- Base salary: USD 180,000 to 350,000 per year
- Equity
- Relocation support
Location and work model
- San Jose, California, United States
- On-site, 5 days per week in office
- Full-time