Shubham Laddha

About Shubham Laddha

ML/AI Accelerator Compiler Engineer | 8+ Years Experience Strategic technical leader specializing in Apache TVM and high-performance deployment of deep learning models on custom silicon. Proven track record at Marvell Technology driving end-to-end compiler optimizations—from graph-level IR transformations to backend code generation for production accelerators.Expert in deploying complex ML/DL workloads on custom accelerators, delivering up to 3× performance improvements on critical operators such as Conv2D and GEMM through intrinsic authoring, low-level scheduling, and architecture-aware optimizations.Strong experience in bridging software and hardware by engineering efficient memory layouts, operator fusion strategies, and data-type quantization pipelines (INT8 / FP16) to achieve scalable, high-throughput inference.

Experience

1. Senior Staff Engineer

   Marvell Technology

   May 2025 — Present · Bengaluru, IN

   Developed and optimized custom TVM compiler passes, including operator fusion and quantization, to accelerate ML models on hardware accelerators.Implemented low-level memory and computation optimizations to reduce usage and maximize throughput.Engineered graph optimizations and high-performance kernels for optimal efficiency.

Education

• Malaviya National Institute of Technology Jaipur

  Bachelor’s Degree, Electronics and Communications Engineering

  2014 — 2018

Skills

• Iot
• Strategic Planning
• Html
• Php
• Data Structures
• Css
• Management
• Verilog
• Public Speaking
• Javascript
• Bash
• Java
• Arduino
• Linux
• C
• Microsoft Office
• Leadership