Ankur Shukla

About Ankur Shukla

As a software engineer specializing in AI hardware acceleration, I work at the intersection of machine learning and systems engineering to optimize neural network performance on cutting-edge AI-specific hardware. My experience spans both training and inference chips across data center and edge deployments, giving me a broad perspective on performance-critical workloads in real-world environments.I have extensive experience with PyTorch, TensorFlow, and ONNX, and have contributed to graph compilers, intermediate representation (IR) transformations, and operator kernel development. I’ve worked on compiler stacks similar to XLA, MLIR, TVM, and Glow, targeting GPUs, TPUs, and custom ASICs/NPU architectures. My work focuses on performance tuning, memory optimization, latency reduction, and hardware-aware model compilation.Key areas of expertise include deep learning compiler stacks, AI chip performance profiling and bottleneck analysis, model quantization, operator fusion, and graph-level optimizations.Hardware Accelerators/GPUs/DSPs for High Computational Intensive AI Workloads in Data Centers (Training and Inference) on applications like Image Recognition/Detection, Recommendation Systems, Diffusion, LLMs, NLP, Video, Generative AI Models. I’m passionate about solving complex performance bottlenecks, collaborating across hardware and software teams, and pushing the limits of what’s possible with modern AI hardware.

Experience

1. Staff Engineer

   Qualcomm

   Jan 2026 — Present · Bengaluru, IN

Education

• NIT SRINAGAR

  Bachelor of Technology (B.Tech.), Electrical, Electronics and Communications Engineering

  2010 — 2014

• Indian Institute of Science (IISc)

  PG Level Advanced Certification Program In Deep Learning, Artificial Intelligence