About Marvell
Marvell’s semiconductor solutions are the essential building blocks of the data infrastructure that connects our world. Across enterprise, cloud and AI, and carrier architectures, our innovative technology is enabling new possibilities.
At Marvell, you can affect the arc of individual lives, lift the trajectory of entire industries, and fuel the transformative potential of tomorrow. For those looking to make their mark on purposeful and enduring innovation, above and beyond fleeting trends, Marvell is a place to thrive, learn, and lead.
Your Team, Your Impact
Marvell Central CAD Engineering is building a new AI-integrated verification design environment, and this role sits at its core. A deterministic Python framework owns everything that decides to pass/fail and everything that must be reproducible — build, run, verdict, coverage, and the quality gates — while an AI layer sits on top for the judgment-heavy work: deciding what to run, triaging failures, and proposing fixes that a human approves. The rule is straightforward: AI proposes, the deterministic framework disposes, a human approves. You will design and own the Python components at the heart of that framework and put them in the hands of the DV engineers who depend on them every day. It is hands-on, high-ownership work with a short path from your code to real impact — the engineers you are building for sit right next to you.
What You Can Expect
Design and own core Python framework components: the declarative build graph and its importer, the run-record store that makes every run reproducible, coverage merge, and verdict logic
Build the simulator backend abstraction — command generation and capability modeling for Cadence Xcelium (MSIE incremental elaboration) and Synopsys VCS — so adding a simulator is a new backend and nothing else changes
Assemble self-contained, token-efficient failure bundles (waveforms, logs, run-record fields, testbench configuration, and source pointers) so downstream agents can debug in one place
Wire the integrations: compute-grid job submission, the results dashboard, CI for the gate-blocking changelist path, and the MCP endpoints the agents consume
Support the AI layer without owning any ML: author reusable agent skills and prompts, build evaluation harnesses to measure triage and fix quality, and enforce the deterministic guardrails around the agents
Package, deploy, and operate the flow — roll it out to verification teams across sites, then monitor and troubleshoot in production
Write the docs and reusable procedures that let the rest of the org adopt the flow