Performance Engineer, Kernelstal by grapevine
Indiranagar, Bengaluru · hybrid · 5-10 yrs · 30-100 LPA
As a Performance Engineer focusing on kernels, you will lead the efforts to push the boundaries of computational efficiency in our machine learning systems. You will work at the intersection of AI hardware acceleration and production-grade software to ensure our agents operate with minimal latency. This role is critical for scaling the underlying infrastructure that powers our talent matchmaking algorithms.
what you'll do:
- Design and implement high-performance CUDA kernels for novel machine learning architectures.
- Optimize existing compute-heavy operations to exceed current industry benchmarks on production hardware.
- Profile and analyze system bottlenecks to reduce inference latency.
- Collaborate with the ML research team to translate research prototypes into efficient, deployable code.
- Maintain high-quality standards for kernel library documentation and testing.
- Participate in the full software development lifecycle from initial design to production deployment.
why join: Tal is building a transformative AI talent agent, providing a unique opportunity to apply extreme optimization techniques to complex machine learning problems that directly impact the future of work.
message me directly on tal BOSS.
tal saysbuild and optimize high-performance ai inference engines and custom cuda kernels for tal by grapevine's talent intelligence platform.
engineering
senior
hybrid
Bangalore
cudac++pythonpytorch
comments 1 anon