Description
You will join Synopsys' TCAD team, which builds simulation engines that power semiconductor process R&D at leading manufacturers worldwide. Our team drives performance breakthroughs by building agentic AI/ML systems that continuously learn to generate multi-physics models used in semiconductor manufacturing.
As a skilled and motivated software engineer, you will design and operate agent-based AI/ML systems that continuously improve how multi-physics models are generated, evaluated, and optimized. You will integrate simulation and execution backends through secure, well-typed tool interfaces, enabling reliable and automated model execution on HPC and containerized runtimes.
Your responsibilities will include establishing closed feedback loops that collect logs, metrics, numerical data, and visual outputs for evaluation and iterative improvement. You will normalize and structure heterogeneous outputs to support downstream analysis and learning across iterations.
You will work at the intersection of AI/ML, explainability, numerical methods, and high-performance computing. You will collaborate closely with physics, TCAD, and software experts to translate ambiguous physical objectives into robust, automated modeling workflows.
To succeed in this role, you will need a PhD or MS in Computer Science, Engineering or Applied Mathematics with 3+ years of experience developing large, complex engineering applications. You will require strong programming skills in Python and modern C++, including work in performance-critical or scientific codebases.
You will also need the ability to design, fine-tune, and deploy domain-adapted LLMs in air-gapped or on-prem environments. You will need to translate ambiguous goals into robust AI architectures that work under real constraints.
Additionally, you will need a deep understanding of agentic systems: planning, tool use, memory, multi-agent coordination, self-improvement loops. You will need hands-on experience with RAG architectures (data ingestion, indexing, retrieval, evaluation).
You will also need proven capability to design and implement MCP servers, exposing secure, well-typed, auditable AI-tool interfaces. Good interpersonal and communication skills are required.