Jaydeep S. — Lead ML Engineer

I am an ML Framework Engineer at Qualcomm, working on building and optimizing high-performance AI systems for custom hardware accelerators. I hold an M.Tech from IIT Hyderabad and specialize in the intersection of machine learning and systems, with a focus on efficient inference and scalable AI infrastructure. At Qualcomm, I work on enabling PyTorch for custom AI accelerators, driving end-to-end execution in eager mode and optimizing operator performance for real-world deployment constraints. My work involves improving latency, throughput, and memory efficiency through hardware-aware optimizations. What I work on: • PyTorch backend integration and operator execution pipeline for custom hardware • Operator optimization (Conv2D/Conv3D, MatMul) using GEMM-based designs and tiling strategies • Latency and throughput optimization under memory and bandwidth constraints • SIMD (HVX) vectorization and multithreaded parallel execution • Efficient utilization of on-chip memory (VTCM) to reduce off-chip bandwidth overhead • Exploration of transformer inference optimizations (KV caching, batching strategies) Core strengths: • Strong foundation in Data Structures & Algorithms and system design • Deep understanding of computer architecture, memory hierarchy, and performance optimization • Experience in profiling, debugging, and optimizing compute-intensive workloads I am particularly interested in ML systems, inference optimization, and large-scale AI serving infrastructure. I enjoy working on complex performance-critical systems and continuously exploring ways to make AI models faster, more efficient, and production-ready. Open to discussions, collaborations, and opportunities in ML systems, AI infrastructure, and performance engineering.

Stackforce AI infers this person is a Machine Learning Engineer specializing in AI infrastructure and performance optimization.

Location: Bengaluru, Karnataka, India

Experience: 4 yrs 1 mo