Portal
Build the cloud platform connecting global Mujin Controller deployments: Go services, remote access, data synchronization, and integrated support.
Team Overview
The Portal team builds the cloud platform that connects and manages Mujin Controllers and robotic systems deployed around the world. Controllers perform real-time computation and robot control at the edge; Portal provides centralized monitoring, management, remote operations, and support.
Managing a few Controllers individually is possible. Managing hundreds or thousands across factories and regions requires shared infrastructure. Portal gives customers and Mujin engineers visibility into Controller status, data, logs, and support issues so they can operate and troubleshoot deployed systems as the fleet grows.
Capabilities include Controller registration, browser-based remote access, data synchronization and backups, dashboards, Virtual Controllers, customer support and ticketing, and an in-product AI assistant. Portal operates in Japan and the US, with additional regions such as Europe planned.
Products & Applications Powered by This Team
Mujin Portal
Centralized visibility, remote access, and support workflows for deployed robotic systems.
Connected Mujin Controllers
Cloud connectivity, registration, data synchronization, and backups for the edge Controllers that execute robot control.
Tech Stack
Backend & Data
Go and GraphQL; graphql-go builds schemas by reflection from Go types. PostgreSQL with GORM and pgvector, plus Redis, support the data layer.
Frontend & Product Services
TypeScript, React 19, MUI, Apollo Client, TanStack Router, and ECharts. Services include Stripe billing, WorkOS SSO, Portal as an OpenID Connect provider, an MCP server, and an AI agent runtime supporting several LLM providers.
Connectivity & Infrastructure
SSH gateway, yamux and SOCKS5 multiplexed tunnels, and an in-process WireGuard server. Multi-region AWS EKS and on-premises Kubernetes clusters use Helm and ArgoCD; object storage uses S3 and MinIO.
Testing & Observability
Real Mujin Controllers in the Tokyo office and Virtual Controllers running full Controller images as Kubernetes pods for simulation and testing. OpenTelemetry, SigNoz, and Grafana provide observability.
Technical Challenges
Connecting a Global Controller Fleet
Challenge: Keep Controllers reachable on customer factory networks while serving multiple regions as one platform. The writable database is in one region, roughly 150 ms across the Pacific from the other.
Approach: Connect through SSH, multiplexed tunnels, and WireGuard. A hub routes connections across nodes and regions; Controller data synchronizes into sharded cloud databases. Design writes around network round trips and serve reads through regional replicas and caches, accounting for each new region’s data residency requirements.
Impact: Hundreds of customer-site Controllers use one platform across two regions. At the time of the intake, the team reported about three production releases per day and more than 550 releases since May 2026.
Building Integrated Customer Support
Challenge: Bring system context and customer support workflows together as Portal expands beyond visibility and management.
Approach: Build tickets with requesters, assignees, and collaborators linked to sites and Controllers; track site incidents and engineering escalations. An Active Monitoring board lets Support view and acknowledge alarms across connected sites. An in-product AI assistant uses tools to read site documentation and system context.
Impact: A site’s tickets, alarms, Controllers, and history sit side by side, giving Support one place to work across the system context and support workflow.
Key Responsibilities
- Design backend services and customer-facing Portal features.
- Build services connecting the cloud product with deployed Controllers.
- Develop Controller data collection, synchronization, storage, and presentation.
- Optimize architecture for scalability, reliability, and maintainability.
- Own features from technical design through implementation, troubleshooting, and maintenance.
- Write design documents for teammate review before coding, and build with AI coding agents in the daily workflow.
Weekly Engineering Rhythm
- 50% Design and code: Go backend and TypeScript frontend.
- 20% Design documents and code review.
- 15% Testing on real and Virtual Controllers.
- 15% Production operations and support.
Who Will Enjoy and Excel on This Team
You will enjoy this team if you:
- Want cloud software to connect directly to physical robotic systems.
- Enjoy backend and systems problems spanning Controllers and cloud services.
- Want to scale from individual devices to fleet-wide operations.
- Enjoy building customer-facing products.
- Like owning design, implementation, deployment, and troubleshooting.
Who Might Struggle on This Team
- Prefer frontend-only work without engaging with backend systems or data.
- Need significant direction to independently deliver well-defined features.
- Prefer narrow implementation tasks without considering broader system behavior.
- Are uncomfortable raising technical blockers proactively or collaborating across teams.
Join the Team
Qualifications & Growth
- Hands-on backend development in Go.
- Relational databases, PostgreSQL, and data modeling.
- Backend APIs (GraphQL or REST) and system design.
- Independent ownership and delivery of software features or services.
- Strong technical problem-solving and communication.
Full-stack opportunities also call for experience building modern web applications that connect backend APIs and data to frontend interfaces. Kubernetes, cloud infrastructure, TypeScript, and React are helpful additional skills.
- Cloud platforms connected to physical robotic systems and edge devices.
- Scaling connected-device management toward fleet-wide operations.
- Real-world data synchronization, remote operations, and observability.
- Customer-facing products spanning backend engineering, cloud infrastructure, edge computing, and robotics.