Michael Melesse
Founder & Managing Partner at Leo Ventures Digital Group
New York, New York, United States
Summary
🤩
Rockstar🎓
Top SchoolMichael Melesse is a seasoned technology entrepreneur and leader with over three decades of experience architecting AI, cloud, and telecom transformations for governments and global enterprises. As Founder & Managing Partner at Leo Ventures and former Global Practice Lead at Dell, he has built multi-billion-dollar business units, led nationwide Telematics deployments, and holds a U.S. patent for secure edge IoT systems. He blends deep telecom systems engineering—from COMSAT and Ericsson—to strategic cloud and AI platform design, partnering with Intel, AMD and Nvidia to operationalize advanced ML and 5G solutions. An active contributor to GPU-focused open source (helping enable AMD ROCm support in Triton and enhancing AMD’s MIOpen), he brings rare hands-on expertise in mixed-precision ML kernels and backend compiler work. Known for turning complex, regulated infrastructure challenges into scalable product roadmaps, he mentors teams to bridge research, deployment and business outcomes.
9 years of coding experience
10 years of employment as a software developer
Bachelor of Science (BSc) w/Honors Computer Science, Bachelor of Science (BSc) w/Honors Computer Science at Western UniversityMaster of Science (MSc) Engineering & Business Management, Master of Science (MSc) Engineering & Business Management at The George Washington University
English, Amharic
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (17)
artificial-intelligence10
c-language10
python10
compiler-development10
hip10
triton10
roc10
amdgpu10
cprogramming-language10
intermediate9
machine-learning9
multilevel9
cuda9
rnn-model8
n8
Programming languages (8)
Github contributions (5)
triton-lang/triton
Dec 2021 - Nov 2022
Development repository for the Triton language and compiler
Role in this project:
Backend Developer
Backend DeveloperContributions:55 reviews, 50 commits, 41 PRs in 10 months
Contributions summary:Michael primarily contributed to enabling and improving the AMD ROCm backend for the Triton language and compiler. Their work involved significant changes to the Python frontend and C++ backend to allow compilation and execution of kernels on AMD GPUs. The user implemented core functionality, addressed code comments, and integrated features from a fork of the repository. These modifications included implementing core functionalities and fixing the shift-op test.
compilerprogramming-languagecode-generationtriton
ROCm/MIOpen
Sep 2018 - Aug 2019
AMD's Machine Intelligence Library
Role in this project:ML Engineer
ML EngineerContributions:10 commits in 10 months
Contributions summary:Michael primarily contributed to refactoring and enhancing the `miopen` library, an AMD machine intelligence library, with a focus on supporting mixed-precision (FP16) for Batch Normalization (BN) operations, specifically BN_3d and spatial tests. They integrated rocBLAS for RNNs and implemented FP16 support across different RNN components such as vanilla, LSTM, and GRUs. In addition, they addressed memory management issues.
amdcluster-computingmachine-learningblasmachine-intelligence
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.