Think deeply. Build fast. Explore beyond.

I am an AI engineer focused on online learning and exploration in interactive systems.
My current work bridges ranking bandit algorithms and AI-driven product development, pursuing systems that adapt through interaction rather than rely only on static training.

• Ranking Bandits / Online Learning – exploration strategies in dynamic environments
• Contextual Decision-making – bandit feedback, uncertainty, implicit reward modeling
• AI Systems – applied intelligence through reasoning + interaction loops

I am currently studying bandit-based recommendation and ranking models and exploring their real-world deployment in dialogue systems and personalized AI.

A conversational engine that learns user preference signals through feedback using contextual bandits.
It doesn’t just generate responses—it actively explores topics to discover what engages the user.

Hachika is an experimental AI project that explores self-oriented artificial behavior. Instead of being designed purely as a helpful tool, Hachika is imagined as an artificial being that forms its own preferences, goals, and way of relating through experience.

Rank Bandit Lab is a lightweight playground for studying ranking bandits. You can define attraction probabilities per document, plug in different ranking policies, and observe click behavior under a cascade-style user model. The project ships with command-line tools, tutorials, scenarios, and notebooks so you can move quickly between simulation and analysis.

I don’t believe technology is just a tool. It’s an extension of how we think, explore, and reach beyond what we know.
Rather than chasing trends, I pursue systems that learn continuously and grow with humans.

I don’t repeat the known. I explore the unknown.

To build exploration-driven AI systems that push interaction beyond static prompts—
and to contribute to the future of human–AI collaboration through online learning research and product engineering.
Future path: Global AI engineering / Applied AI research (GAFAM or Research Track).

Python / TypeScript / React / FastAPI / Next.js / MongoDB / AWS / Git
(Research) PyTorch / Bandit Algorithms / Recommender Systems

• X (Twitter): https://x.com/YaSut0ra94970
• Portfolio: https://yasut0ra-portfolio.vercel.app/
• LinkedIn: https://www.linkedin.com/in/takuma-yasuda-7a332533b/

北海道大学 情報工学系の学生 / オンライン学習（バンディット）を研究しながら、AIプロダクト開発にも取り組んでいます。
研究と実装の両軸で、人とAIが相互学習するシステムの可能性を追求しています。