Applied Scientist II, Machine Learning Accelerator - Annapurna Labs
Core
Designing a formal methods-based framework using SMT solvers to generate legal and realistic stimulus for custom Machine Learning Accelerator Chip verification.
Role type
Applied Scientist (Hardware Verification & Formal Methods)
Builds
Automated content generation framework for pre-silicon design verification of ML accelerators
Domain
Hardware verification, Formal methods, Machine Learning Accelerators
Deliverable
production ML models | infrastructure
Required skills
First-order logic modeling, SMT solver development, algorithm design, numerical optimization, parallel/distributed computing, high-performance computing, Java/C++/Python programming
Preferred skills
Unix/Linux development, professional software engineering
Technologies
SMT solvers, Java, C++, Python
Responsibilities
Develop understanding of custom ML instruction set architecture, model instruction stream correctness using first-order logic, create custom APIs for scheduling and randomness control, deploy algorithms for concurrent code safety, create coverage metrics, use machine learning to automate content creation
Seniority
Mid-level, hands-on IC
