Tsubaki๐Ÿ”—

Building quieter, more efficient ways for machines to learn.

I am Tsubaki, a software engineer and independent researcher exploring the space between brain-inspired computing, neuromorphic systems, and practical AI infrastructure.

Read the blog ยท View my GitHub ยท Get in touch


What Iโ€™m exploring๐Ÿ”—

Neuromorphic intelligence๐Ÿ”—

Spiking neural networks, memory-based models, and architectures that can learn with less data, energy, and latency.

Systems that stay out of the way๐Ÿ”—

Reliable software for real-time workloads, from low-level networking and concurrency to efficient inference pipelines.

Research made tangible๐Ÿ”—

I like turning ambitious ideas into small, testable systems โ€” preferably in Rust, with clear interfaces and measurable behavior.

Selected work๐Ÿ”—

IrohaLLM๐Ÿ”—

An exploration of memory-based language models and their first applications to spiking neural networks.

LLMs ยท SNNs ยท memory systems

Around here๐Ÿ”—

The blog is where I write about the engineering details behind the ideas: architecture notes, experiments, networking, and the occasional reverse-engineering session.

If you are working on efficient AI, unusual computer architecture, or dependable real-time systems, I would be glad to hear from you.


Find me elsewhere: GitHub ยท Email

GPG fingerprint: 33D3 ECCF B9D8 0D00 0721