Abhilash Majumder
Bengaluru, Karnataka, India
Summary
🤩
Rockstar🎓
Top SchoolAbhilash Majumder is an AI compilers and frameworks engineer with 8 years of experience building high-performance ML tooling for Intel and Habana, after stints at Morgan Stanley and HSBC. He specializes in deep learning compilers, SYCL/XPU and GPU backends, and has contributed performance-critical work to flagship open-source projects such as llvm, llama.cpp, whisper.cpp and Hugging Face’s accelerate. His contributions span enabling XPU/IPEX support, SYCL backend ports, novel quantization methods and NVPTX intrinsics—bridging low-level compiler internals with practical LLM training and inference optimizations. Based in Bengaluru and trained at NIT Durgapur, he combines production-grade engineering with a habit of upstreaming hardware-aware optimizations that accelerate real-world model deployments.
8 years of coding experience
Bachelor of Technology Computer Science, Bachelor of Technology Computer Science at National Institute of Technology Durgapur
Stackoverflow
124reputation
79kreached
7answers
0questions
Github Skills (50)
pytorch10
c-language10
intrinsics10
distributed-training10
python10
algebra10
llvm10
llama10
whisper10
gpu-programming10
machine-learning10
inference10
cmake10
sdp10
nvtx10
Programming languages (12)
Github contributions (5)
ggml-org/llama.cpp
Aug 2023 - Mar 2025
LLM inference in C/C++
Role in this project:
Back-end Engineer (SYCL/GPU Optimization)
Back-end Engineer (SYCL/GPU Optimization)Contributions:116 reviews, 36 PRs, 29 pushes in 1 year 6 months
Contributions summary:Abhilash primarily contributed to the SYCL backend implementation for Intel GPUs within the llama.cpp repository. Their work involved porting and optimizing code for SYCL, including refactoring and converting CUDA code. They fixed issues related to the f16_sycl and other copy calls, implemented new quantization methods (q3_s and q1_s), and improved performance through macro usage. Their contributions are critical for enabling GPU acceleration for LLM inference.
ggmlllama
meta-llama/llama-cookbook
Aug 2023 - Apr 2024
Welcome to the Llama Cookbook! This is your go to guide for Building with Llama: Getting started with Inference, Fine-Tuning, RAG. We also show you how to solve end to end problems using Llama model family and using them on various provider services
Role in this project:ML Engineer
ML EngineerContributions:3 reviews, 3 PRs, 6 comments in 8 months
Contributions summary:Abhilash's commits primarily focus on enabling and improving the training and inference capabilities of Llama models within the repository. The contributions include adding support for XPU finetuning and inference, fixing data loading bugs, and integrating with Intel's IPEX for performance optimization. They also made memory management improvements, including memory tracing, and resolved merge conflicts while also adding the ability to enable grad on loss tensor. Their work directly impacts the efficiency and platform compatibility of running and fine-tuning Llama models.
aifinetuninglangchainllamallama2
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.