Senior Performance Engineer, Intel Stack
Sarvam AI
Bengaluru, IndiaPosted 3 months ago
S
Skill Required
Infrastructure AssemblybuildingCIAIMachine LearningFulltime
Key highlights
- 5+ years of machine learning deployment experience, including 2+ years on Intel inference stacks.
- Production OpenVINO experience with model conversion, accuracy validation post‑quantization, and driver‑version pinning.
- High ownership and high impact role from day one.
- Opportunity to work on AI problems that could affect an entire country.
- Work alongside a fast‑moving, high‑talent team backed by notable investors (Lightspeed, Peak XV, Khosla Ventures).
- Direct interaction with Intel’s ecosystem and OpenVINO team as the technical liaison.
Role overview
The role is responsible for owning Sarvam's Intel surface end‑to‑end, including Intel NPU via OpenVINO (Meteor Lake, Lunar Lake, vPro AI PCs), integrated and discrete GPUs via OpenVINO and ONNX Runtime, and x86/AMD64 CPU optimization for fallback paths. The engineer will serve as the technical liaison to Intel’s ecosystem and the OpenVINO team, ensuring models meet defined SLAs and managing the device‑CI pipeline.
Responsibilities
- Own Sarvam's Intel surface end‑to‑end: Intel NPU via OpenVINO, Intel integrated and discrete GPU via OpenVINO and ONNX Runtime, and x86/AMD64 CPU optimization for fallback paths.
- Be the technical face to Intel's ecosystem and the OpenVINO team.
- Land Sarvam’s edge models on Intel NPU / dGPU / iGPU inside defined SLAs.
- Own the OpenVINO build/quantization recipe for the team's models, including the driver‑version compatibility matrix.
- Drive x86 CPU optimization for the universal fallback path – AVX‑512, AMX, threading strategy.
- Own the Intel device‑CI pool and regression detection across OpenVINO upgrades.
Requirements
- 5+ years on ML deployment with 2+ years on Intel inference stacks.
- Production OpenVINO experience including model conversion, accuracy validation post‑quantization, and driver‑version pinning.
- ONNX Runtime EP knowledge – when to use OpenVINO EP vs. native OpenVINO, and when to fall back to CPU EP.
- x86 CPU profiling and optimization – VTune, perf, comfort reading hot loops at the assembly level when needed.
Nice to have
- AVX‑512 or AMX intrinsics is a strong plus, not a must‑have.
- Direct prior interaction with the OpenVINO team or Intel ecosystem partners.
- Custom OpenVINO operator authoring.
Benefits
- Fast‑moving, high talent‑density team building full‑stack AI for India.
- Work alongside researchers, engineers, builders, and business leaders who move fast and hold each other to a very high bar.
- High ownership and high impact from day one.
- AI‑first approach in building, shipping, and problem solving.
- Opportunity to work on problems that could change how an entire country learns, works, and communicates.
- Ability to work on frontier AI problems in India.
Additional details
- Sarvam is building the bedrock of Sovereign AI for India, developing a full‑stack sovereign AI platform across research, models, infrastructure, and applications with a focus on making AI genuinely work for India.
- Sarvam works with leading enterprises and public institutions and is backed by Lightspeed, Peak XV, and Khosla Ventures.
- Sarvam partners with India’s leading brands, including Tata Capital, SBI Life, CRED, IDFC, and LIC.