We are seeking an exceptional Kernel Developer with a strong background in building low-level, performance-critical software for cutting-edge AI and ultra-low-latency compute platforms. You will collaborate closely with hardware, architecture, and compiler teams to design and optimize deep-learning kernels for next-generation processors.

Key Responsibilities :

- Design, implement, and optimize low-level kernels for AI deep-learning algorithms targeting ultra-low-latency processors.

- Collaborate with hardware engineers, architects, and compiler teams to integrate software with novel hardware features.

- Perform profiling, benchmarking, and tuning of kernels to ensure optimal performance, resource utilization, and minimal latency.

- Develop clean, maintainable, and efficient code in C/C++ and Python, contributing to internal libraries and tooling.

- Analyze complex algorithms and large codebases to identify bottlenecks and propose high-impact optimizations.

Requirements :

- Bachelors or Masters degree from Top Premium Colleges such as IITs, NITs, IIITs, BITS Pilani, IISc, or equivalent Tier-1 institutions.

- Strong proficiency in C/C++ programming, especially in performance-critical environments.

- Proven experience in developing high-performance kernels for GPUs or custom accelerators.

- Deep expertise in optimization techniques : vectorization, memory hierarchies, parallelization, cache utilization, etc.

- Strong ability to understand and enhance complex codebases and algorithms.

- Excellent debugging, profiling, and problem-solving skills.

- Strong communication and teamwork skills, with the ability to work across hardware and software boundaries.

Preferred / Added Advantage :

- Understanding of microprocessor instruction set architectures (ISA).

- Experience with Python-based kernel development using Triton or similar frameworks.

- Experience working with distributed systems and communication protocols like NCCL, MPI, Libfabric.

- Knowledge of neural network architectures, quantization, and precision optimization.