01 | ABOUT
Freelance software engineer with commercial experience delivering work for companies. I write systems-level code in C and C++, build machine learning tooling in Python and CUDA, and care about software that is small, readable and easy to audit.
My approach is simple: less code means fewer bugs, a smaller attack surface and a faster review. I prefer minimal dependencies, POSIX-compatible tooling and designs that one person can read and fully understand in an afternoon.
I also contribute to open source projects when I can be useful.
02 | FOCUS
- Systems programming: C and C++, POSIX, low-level tooling, performance work
- Machine learning engineering: model quantization, QLoRA fine-tuning, inference optimization
- GPU computing: CUDA kernels and GPU-accelerated pipelines
- Security and auditing: minimal, reviewable code with a small trusted base
- Alternative operating systems: Darwin, FreeBSD, OpenBSD, NetBSD, Plan 9, illumos
- Terminal-centric workflow: Neovim, tmux, shell tooling, suckless-style software
03 | LANGUAGES
Fluent in C, C++ and Python, with professional experience in eight more languages.
Also writing Holy C, Pascal and 1C.
04 | OPERATING SYSTEMS
I run and target a range of systems, with a particular interest in Darwin internals and the BSD family.
05 | MACHINE LEARNING
Hands-on engineering experience in applied AI and ML, with emphasis on efficient training and inference on constrained hardware.
- Quantization of large language models for lower memory use and faster inference
- QLoRA fine-tuning and other parameter-efficient methods
- CUDA development for custom GPU workloads
- Model publishing and maintenance on Hugging Face
Hugging Face: huggingface.co/tinyopsec
06 | TOOLING
Daily work happens in the terminal, with tmux, Make and suckless-style software.
07 | PRINCIPLES
- Simplicity. The simplest design that solves the problem is the right one.
- Auditability. Code should be short enough to read in full and clear enough to verify.
- Minimalism. Every dependency and every feature has a cost, so each one must justify itself.
- Performance. Measure first, then optimize what matters.
- Learning by building. Real projects teach more than tutorials.
08 | WORK WITH ME
Available for freelance engagements in:
- Systems and low-level development in C and C++
- Machine learning engineering: quantization, fine-tuning, deployment
- CUDA and performance optimization
- Code review and security-minded refactoring
- Tooling and automation
Contact: open an issue or discussion on any repository, or reach out through the links on my GitHub profile.
09 | STATISTICS
