Kai Hirota
Software Engineer @ Stake · Co-Founder @ Koi Labs
Software engineer from Japan interested in distributed systems, knowledge representation, and NLP.
Experience
Immutable
Software Engineer II, Core EngineeringSydney, Australia
Aug 2022 - Jan 2025
Immutable is a global leader in gaming on a mission to bring digital ownership to every player by making it safe and easy to build great web3 games.
- Go
- TypeScript
- PostgreSQL
- Redis
- AWS
- Terraform
- Docker
- Kubernetes
- Monolith decomposition. Drove core workstreams in a year-long monolith-to-microservices decomposition, designing zero-downtime rollout sequencing and idempotent consumer patterns to safely shift from exactly-once to at-least-once event delivery for high-throughput financial transactions.
- Webhook platform. Built customer-facing webhook delivery with cross-team collaboration; payloads matched API response schemas with SDK deserialization and signature validation, plus self-service delivery logs and success-rate monitoring.
- Asset migration. Designed a migration service for ~10M digital assets with a combined market cap of $20–30M between two storage systems with incompatible schemas.
Mojexa
Software Engineer InternCanberra, Australia
Feb 2022 - Jun 2022
- Unreal Engine 5
- C++
- AWS Kinesis
- Redis Pub/Sub
- WebSockets
- Real-time IoT pipeline. Built a real-time backend ingesting IoT sensor data into Kinesis, fanning out via Redis Pub/Sub to WebSocket servers that push location updates to subscribed Unreal Engine clients.
- UE5 C++ plugin. Bound in-game actors to real-world entities, with coordinate transforms for GPS-to-game mapping and support for both live tracking and historical replay.
Airbnb
Data Scientist InternTokyo, Japan
Aug 2019 - Jan 2020
- SQL
- Apache Superset
- Jupyter Notebook
- Pandas
- GeoPandas
- QGIS
- Mapbox
- Python
- Tokyo 2020 strategy. Partnered with executives to deliver data-driven insights informing business strategy for the Tokyo 2020 Olympics, translating statistical analyses into actionable recommendations for non-technical stakeholders.
- Kyoto siting analysis. Determined optimal check-in station locations in Kyoto using geospatial analysis and Poisson-modeled arrival patterns, minimizing cost while satisfying legal distance requirements and peak capacity constraints.
Fracta.ai
Data EngineerRedwood City, CA
Oct 2018 - Apr 2019
Fracta uses machine learning to predict failure of water pipes.
- SQL
- Jupyter Notebook
- Pandas
- GeoPandas
- QGIS
- Mapbox
- Python
- OSGeo
- Shapely
- Geospatial preprocessing. Built an automated pipeline ingesting heterogeneous vector and raster inputs (pipe networks, soil, traffic, elevation) and transforming them into features for ML prediction of water pipe failure likelihood.
- Vector graph reconstruction. Developed a vector data cleaning system that reconstructs connected network graphs from noisy pipe geometry — inferring node positions and connectivity from raw edge coordinates without known topology.
Education
Australian National University
Master of Computing (Machine Learning)Canberra ACT, Australia
Mar 2021 - Aug 2022
- ENGN6528 Computer Vision
- COMP8600 Statistical Machine Learning
- COMP6490 Natural Language Processing
- COMP6300 Operating System
- COMP6442 Software Construction
- COMP6331 Computer Networks
University of Sydney
Master of Data ScienceSydney, Australia
Mar 2020 - Jan 2021
- Probability
- Statistical Data Analysis
- Algorithms
- Database Management Systems
- Software Engineering in Java
- Machine Learning
- Information Theory and Self-Organization
- Advanced Data Models
Santa Clara University
Bachelor of Commerce (Information Systems & Analytics)Santa Clara, California, USA
Oct 2016 - Oct 2018
- Introduction to Programming in C
- Systems Programming
- Operation Management
- Strategic Analysis Capstone
- Statistics & Data Analysis