Perception
Computer vision pipelines, representation learning, and deployment-minded evaluation.
Machine Learning Engineer in Tokyo
Rinat builds ML systems that learn from messy, real-world signals.
I work across computer vision, NLP, reinforcement learning, data analytics, and machine learning competitions, turning research ideas into practical systems.
Computer Vision NLP RL Analytics Competitions
Tokyo building applied ML
CV + NLP + RL research to production
Competitions ranking pressure, clean validation
PyTorch stack experiments, metrics, iteration
What I Like Working On
Models that get sharper under feedback.
I am interested in the loop where data quality, model behavior, and product constraints collide. That includes visual perception, language models, RL-style optimization, and the practical details that make experiments reproducible.
Language
NLP systems, retrieval, embeddings, and text classification for real documents.
Learning Loops
Optimization, RL-inspired workflows, analytics, and fast competition iteration.
Recent Notes
Short logs from the ML workbench.
Modern ML Practice in 2026
This is a short snapshot of practical machine learning work as of 2026. The biggest shift is that many projects now start from ...
Regularization
Regularization is any training choice that helps a model generalize instead of only memorizing the training set. It can be a pe...
Embeddings
Machine learning models do not understand raw text directly. They need text to be converted into numeric vectors. An embedding ...
I am also part of the Open Data Science community: a good place for ML practice, competitions, and hard-won debugging instincts.