Design and build benchmark suites covering inference performance, model quality, and knowledge evaluation across hardware targets.Run external partner verifications, compare solutions against benchmarks, identify gaps, and clearly communicate findings.Port models such as LFM2 across runtimes and frameworks and verify correctness end-to-end.Maintain and extend the inference engine la
