whoami - Martin's website/blog thingy

Who Am I?

I'm Martin, a software engineer based in Taiwan specializing in high-performance software, AI computing, low-level engineering, and Linux systems. Now exploring OpenBSD and privacy-related software development.

I don't really stick to one stack, but if I had to choose, my favorites are:

Contact

Feel free to reach out if you want to know more or have questions. I'm always happy to chat about anything technical.

Email

martin at clehaxzetw dot tw

Jami

QR code (P2P, encrypted)

PGP

1ABCDF0F8548C3B9D

Matrix

@clehaxze:matrix.clehaxze.tw

I'm open to contracted work. If interested, here's my resume.

Languages

Mandarin - Native
English - Fluent
Esperanto - Moderate

Work Experience

AINekko

Sr. Platform Engineer

2026-01 ~ Present

Built out the operator library, contribute to next gen ISA desies, working with truly smart people and keep running the missing of open source AI to improve all our lives.

Tenstorrent

Sr. Staff Developer Relations Engineer

2025-04 ~ 2026-01

Writing excellent documentation, tutorials, and growing the community around Tenstorrent's AI accelerators.

SatLayer

Senior SDK Engineer

2024-08 ~ 2025-04

Founded by one of Lumina's founders. Invited back to manage and develop SDK for DX and other projects.

NVIDIA

Senior Software Engineer

2023-11 ~ 2024-08

Worked in the Omniverse team on platform development, software integration, and cross-domain programming.

Lumina Industries, Inc.

Director of Product Engineering

2022-10 ~ 2023-11

Official title for signing product packages. Still a software engineer at heart.

Native Application Lead

2022-01 ~ 2022-10

Architected, developed, and coordinated the company's native application, including image pipeline, UI, and AI systems.

C++ Developer

2021-09 ~ 2022-01

Developed company applications and algorithms. Introduced the team to modern C++ standards and tools like Address Sanitizer and FBInfer.

Education

National Sun Yat-Sen University

Computer Science and Engineering

2016 ~ 2021

Graduation project: Hierarchical Temporal Memory Agent in standard Reinforcement Learning Environment

Research Assistant

2016-12 ~ 2021-06

Besides learning. I also participate in research and co-development of machine learning accelerators, compilers for model execution, among various research projects.

Major Open Source Projects

llama.cpp

Arguably the most popular LLM inference engine for the open-source community. I wrote the RK3588 and Tenstorrent forks, sparked the RK3588 NPU reverse engineering effort, gave a FOSDEM talk, and am working on upstreaming the Tenstorrent port.

drogon

Production-ready, extremely fast C++ web framework. I'm a maintainer - wrote the coroutine subsystem, HTTP/2 client, and various bug fixes. Easily handles C10K on embedded systems and C100K on standard servers.

Etaler

High-performance implementation of Hierarchical Temporal Memory, a biologically-inspired machine learning model developed by Numenta. At release, 20x+ faster than HTM.core on CPU and another 2x on GPU.

embree-arm

World's first functional port of Intel's Embree ray tracing kernels for ARM processors. Though no longer maintained, this project inspired embree-aarch64 using the same porting approach.

Conference Talks

I'm on a biannual schedule giving talks at public conferences on topics I've researched and feel confident about. Speaker at SITCON (Student's Information Technology Conference) and COSCUP (Conference for Open Source Coders, Users, and Promoters).