Data Science • Infrastructure • Technical Writing
データサイエンス・インフラストラクチャ・技術文書
Portfolio & Research

Lite analytics, machine learning, and data visualization tools running on disposable infrastructure

Kubernetes
Elasticsearch
JupyterHub

Platform Services
プラットフォームサービス

Azure
k3s k3s
Elastic Elastic
Superset Superset
Jupyter Jupyter
NVIDIA NVIDIA
Meshtastic
Translation

Rancher

Kubernetes cluster management platform for container orchestration, monitoring, and multi-cluster operations

K8s Management Multi-Cluster GitOps
Open Rancher

Container Registry

Self-hosted OCI-compliant container registry powered by Zot, with Azure Container Registry sync and Keycloak OIDC authentication

OCI ACR Sync OIDC
Open Registry

Git

Self-hosted Git service powered by Gitea for source code management, CI/CD integration, and private repository hosting

Gitea CI/CD Private Repos
Open Gitea

Elasticsearch

Advanced search and analytics engine with Kibana for real-time data visualization and insights

Real-time Scalable Analytics
Open Kibana

JupyterHub

Multi-user data science environment with GPU support, persistent storage, and collaborative notebooks

Python & R GPU Enabled
Launch JupyterHub

Apache Superset

Operator-facing BI and dashboard platform visualizing Meshtastic mesh activity, GNSS tracks, signal quality, and sensor telemetry from Elasticsearch

Dashboards Geo Maps Keycloak SSO
Open Superset

Apache NiFi

Visual data flow automation platform ingesting MQTT telemetry, transforming Meshtastic packets, and routing data into Elasticsearch

Data Pipelines MQTT Ingest Keycloak SSO
Open NiFi

Meshtastic

LoRa mesh radio network daemon running on the control plane, bridging field nodes to the platform via MQTT for DDIL-resilient communications

LoRa Mesh DDIL GPS Tracking

Mosquitto MQTT

Lightweight MQTT broker receiving telemetry, position, and message data from Meshtastic mesh nodes and routing it to the NiFi pipeline

MQTT Real-time Mesh Bridge

Sensor & Recognition

Object recognition camera and sensor integration for real-time detection, classification, and environmental data collection

Object Detection Sensors Real-time

Monitoring

Datadog observability platform deployed via Helm operator for comprehensive infrastructure and application monitoring

Datadog Helm Operator APM

ML Pipeline

End-to-end machine learning pipeline with model training, deployment, and inference capabilities

AutoML Deployment APIs

Azure Cloud

Microsoft Azure serving as the cloud backbone — hosting ACR as the upstream image registry, Velero backups via Storage Account, and Azure Arc for unified cluster management across edge deployments

ACR Velero Backups Azure Arc
Azure Portal

Edge Translation

On-device speech-to-text and multilingual translation powered by the NVIDIA Jetson Orin Nano — converting voice to text and broadcasting in English, Japanese, Spanish, and Mandarin over the mesh with zero internet dependency

Off-Grid Whisper STT NLLB-200

Single Sign-On

Keycloak identity platform providing unified authentication across all cluster services via Google SSO and local DDIL accounts for offline operations

Keycloak Google SSO Federated
Open Keycloak

Upcoming Projects 今後のプロジェクト

Defense Research Lab

Technical Defense & Security Analysis

Expanding technical defense capabilities with dedicated research infrastructure for vulnerability analysis, threat modeling, and defensive security research.

  • Vulnerability scanning automation
  • Malware analysis sandbox
  • Network traffic analysis
  • Threat intelligence integration

Multi-Cloud Integration

Azure, AWS, GCP Hybrid Architecture

Building hybrid cloud infrastructure spanning Azure, AWS, and GCP for resilience, cost optimization, and multi-region deployment.

  • Cross-cloud Kubernetes federation
  • Multi-cloud data synchronization
  • Unified monitoring & logging
  • Cost optimization strategies

Portfolio ポートフォリオ

Showcasing technical projects, infrastructure implementations, and data science work across cloud platforms, machine learning, and security research.

Data Science

Apple Health Data Analytics

Built automated pipeline to parse, transform, and visualize Apple Health data in Elasticsearch. Created custom Kibana dashboards for workout tracking, heart rate analysis, and health metrics visualization.

Python Elasticsearch Kibana Data Visualization
View Project
Infrastructure

Kubernetes Homelab Cluster

Designed and deployed production-grade Kubernetes cluster with JupyterHub, Elasticsearch, Keycloak authentication, persistent NFS storage, and Let's Encrypt TLS automation via Traefik.

Kubernetes Docker Helm Traefik
Security

OAuth2 Security Research

Deep-dive analysis of OAuth2 Proxy, Keycloak integration, and Elasticsearch authentication patterns. Documented licensing limitations, security trade-offs, and best practices for enterprise authentication.

OAuth 2.0 Keycloak Security OIDC

More Projects Coming Soon

Additional portfolio items will be added as projects are completed and documented.

Technical Defense Blog 技術防衛ブログ

Technical analysis, security research, and defensive strategies for modern infrastructure and applications.

OAuth2 vs Native OIDC: Elasticsearch Authentication Deep Dive

Analysis of authentication patterns for Elasticsearch, comparing OAuth2 Proxy implementation with native OIDC integration. Covers licensing limitations, security implications, and practical deployment considerations.

Read Article

Building a Production Kubernetes Cluster at Home

Complete guide to deploying enterprise-grade Kubernetes infrastructure on homelab hardware, including storage, networking, and security best practices.

Coming Soon

Edge AI with NVIDIA Jetson: Real-time Computer Vision

Implementing computer vision models on Jetson Orin Nano for real-time inference, including DeepStream pipelines and Kubernetes integration.

Coming Soon

Subscribe for Updates

New technical articles and defense research published regularly. Topics include security, infrastructure, and data science.

Built for Scale スケーラブルな設計

Our data science platform runs on a modern Kubernetes infrastructure, providing enterprise-grade reliability and scalability. From data ingestion to model deployment, every component is designed for performance.

Kubernetes
Docker
Python
GPU Compute
Azure
Linux
User Interface
Microservices
Kubernetes
Infrastructure