Henri Vasserman
Senior Developer at Bondora
Tallinn, Harju maakond, Estonia
Summary
🤩
Rockstar🎓
Top SchoolHenri Vasserman is a Senior Developer based in Tallinn with 11 years of professional experience building web and backend systems, currently driving engineering work at Bondora. He brings deep practical knowledge from a long tenure at MarkIT.eu into scalable product development and ongoing delivery in fintech. Henri also contributes to cutting-edge open-source ML tooling—most notably adding CUDA/ROCm GPU acceleration, non-contiguous tensor support, and OpenLLaMA 3B improvements to the widely used llama.cpp project—demonstrating low-level performance tuning skills beyond typical backend work. With a BSc in Information Technology from TalTech, he pairs solid academic foundations with hands-on systems optimization and a knack for squeezing performance from both CPU and GPU stacks.
11 years of coding experience
10 years of employment as a software developer
BSc, Information Technology, BSc, Information Technology at TalTech – Tallinn University of Technology
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (7)
cuda10
llama10
opencl10
c-language10
gm10
gpu-acceleration10
cprogramming-language10
Programming languages (10)
Github contributions (5)
ggml-org/llama.cpp
Mar 2023 - Jan 2025
LLM inference in C/C++
Role in this project:
ML Engineer
ML EngineerContributions:202 reviews, 50 PRs, 69 pushes in 1 year 10 months
Contributions summary:Henri primarily focused on integrating and optimizing code for NVIDIA CUDA and AMD ROCm GPU acceleration within the llama.cpp framework. Their contributions included significant modifications to the `ggml-cuda.cu` and `ggml-cuda.h` files, enabling non-contiguous tensor support and improving data transfer efficiency. They also worked on OpenCL kernel optimization, including fixes for new data formats and older devices. Additionally, the user added support for OpenLLaMA 3B model and increased scratch buffer sizes for it.
ggmlllama
SlyEcho/llama.cpp
Apr 2023 - Aug 2023
Port of Facebook's LLaMA model in C/C++
Contributions:6 reviews, 2 PRs, 183 pushes in 4 months
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.