<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Home on Daisuke Nohara</title><link>https://profile.neodymium6.net/</link><description>Recent content in Home on Daisuke Nohara</description><generator>Hugo -- 0.155.2</generator><language>en-us</language><lastBuildDate>Sat, 11 Jul 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://profile.neodymium6.net/index.xml" rel="self" type="application/rss+xml"/><item><title>AdaptFM @ ICML 2026</title><link>https://profile.neodymium6.net/activities/icml-2026/</link><pubDate>Sat, 11 Jul 2026 00:00:00 +0000</pubDate><guid>https://profile.neodymium6.net/activities/icml-2026/</guid><description>&lt;p&gt;In July 2026, I attended ICML 2026 at COEX in Seoul, South Korea, and presented our paper, &lt;strong&gt;On the Optimal Reasoning Length for RL-Trained Language Models&lt;/strong&gt;, at the &lt;a href="https://adaptfm.gitlab.io/"&gt;AdaptFM workshop&lt;/a&gt;.&lt;/p&gt;
&lt;p&gt;&lt;a href="https://profile.neodymium6.net/posters/adaptfm-icml-2026-optimal-reasoning-length.pdf"&gt;View poster →&lt;/a&gt;&lt;/p&gt;
&lt;figure class="align-center "&gt;
&lt;img loading="lazy" src="https://profile.neodymium6.net/images/activities/icml-2026/poster-presentation.webp#center"
alt="Daisuke Nohara standing beside his research poster at AdaptFM, ICML 2026" width="420"/&gt; &lt;figcaption&gt;
&lt;p&gt;Presenting our work at the AdaptFM poster session.&lt;/p&gt;
&lt;/figcaption&gt;
&lt;/figure&gt;
&lt;figure class="align-center "&gt;
&lt;img loading="lazy" src="https://profile.neodymium6.net/images/activities/icml-2026/icml-2026-venue.webp#center"
alt="Daisuke Nohara at the ICML 2026 venue signage in COEX" width="560"/&gt; &lt;figcaption&gt;
&lt;p&gt;At the ICML 2026 venue in COEX, Seoul.&lt;/p&gt;
&lt;/figcaption&gt;
&lt;/figure&gt;</description></item><item><title>Homelab</title><link>https://profile.neodymium6.net/hobbies/homelab/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://profile.neodymium6.net/hobbies/homelab/</guid><description>&lt;h2 id="overview"&gt;Overview&lt;/h2&gt;
&lt;p&gt;I run a self-hosted homelab on Proxmox VE, fully automated with Infrastructure as Code.
The entire stack—from VM provisioning to service deployment—is managed by Terraform and Ansible, requiring only a single &lt;code&gt;make all&lt;/code&gt; command to deploy from scratch.&lt;/p&gt;
&lt;p&gt;&lt;img alt="Homelab Grafana dashboard" loading="lazy" src="https://profile.neodymium6.net/images/hobbies/homelab/homelab-grafana-dashboard.webp"&gt;&lt;/p&gt;
&lt;p&gt;Source code: &lt;a href="https://github.com/neodymium6/homelab"&gt;[GitHub]&lt;/a&gt;&lt;/p&gt;
&lt;h2 id="architecture"&gt;Architecture&lt;/h2&gt;
&lt;p&gt;The homelab follows a bastion host pattern. My local machine provisions a bastion VM, which then acts as the infrastructure controller for all internal VMs. Internal VMs are not directly accessible from outside—SSH access is restricted to the bastion only.&lt;/p&gt;</description></item><item><title>Reversi</title><link>https://profile.neodymium6.net/hobbies/reversi/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://profile.neodymium6.net/hobbies/reversi/</guid><description>&lt;h2 id="overview"&gt;Overview&lt;/h2&gt;
&lt;p&gt;A full-stack Reversi (Othello) project built from scratch—from a high-performance Rust game engine to a web app and AlphaZero-style AI training pipeline.
The game engine uses bitboard representation for fast move generation and search. It is published as both a Rust crate and a Python package, and powers the web application and various AI experiments.
As of February 14, 2026, ranked 24th on the &lt;a href="https://www.codingame.com/multiplayer/bot-programming/othello-1/leaderboard"&gt;CodinGame Othello leaderboard&lt;/a&gt;.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Live Demo&lt;/strong&gt;: &lt;a href="https://reversi.neodymium6.net"&gt;Play Reversi in your browser&lt;/a&gt;&lt;/p&gt;</description></item></channel></rss>