Dani Purwadi.

Owning the Kubernetes platform that hundreds of mission-critical banking applications run on, and the Golang tooling around it.

• 01

  Stewarded a high-scale Kubernetes production environment

  Maintained cluster stability and performance across thousands of pods backing hundreds of mission-critical applications.

• 02

  Technical lead for the team's Golang applications

  Guided development, set technical standards, and held the bar on code quality.

• 03

  Introduced Chaos Engineering to monthly DR exercises

  Integrated Chaos Mesh into the monthly “Single-Region Wednesday” disaster-recovery testing.

• 04

  Built L2 validation for custom Operators with Chainsaw

  Automated verification of Kubernetes Operators and CRDs, materially improving deployment reliability.

• 05

  Shipped a CLI for auto-upgrading Helm charts

  Abstracted deployment complexity so developers stopped needing to touch chart internals.

Founded and ran the Cloud Enterprise team from zero, architecting integrations between cleaning robots and the systems they live alongside.

• 01

  Founded and directed the Cloud Enterprise team

  Managed 2 engineers, scaling from the ground up to serve strategic robotics and enterprise partners.

• 02

  Managed the integration portfolio for 20+ partners

  Engineered solutions bridging robot operations with external hardware (lifts, doors) and enterprise systems (CCTV, cloud).

• 03

  Championed code quality and architectural integrity

  Ran technical design sessions and a mandatory peer-review process to keep standards high as the team scaled.

Backend and platform work across a fleet of internet-connected cleaning robots — real-time pipelines, sync, and operator tools.

• 01

  Real-time, high-frequency robot status processing in Go

  Service handles 1,000 concurrent msg/s and autonomously raises and resolves critical alerts. Detection moved from 1–2 hours to <10s.

• 02

  Resilient async file-retrieval system for remote robots

  Queue-based architecture ensures download requests are processed reliably despite intermittent robot connectivity.

• 03

  Remote console for internal users

  Pioneered a feature using node-pty and XTerm so support could reach a robot's OS terminal directly. Cut debug time from 4–5 minutes to <30s.

• 04

  Fixed WebFlux thread-pool misconfiguration on a critical backend

  Service uptime moved from 93.9% → 99.9%; p90 dropped from 30s → ~1s.

• 05

  Revamped the core bi-directional robot ↔ cloud sync

  New Spring Boot system improved sync success from 30% → 95%, cut database load 20× and sync time 10×.

Backend work across the payments platform — flaky tests, e-wallet integration, and third-party reliability.

• 01

  Automated flaky test detection and proactive alerting

  Cut CircleCI infrastructure cost for running tests by 50% (~USD$5,000/month saved).

• 02

  Led DANA e-wallet integration to the Xfers ecosystem

  Connected 115 million DANA users to Xfers' payment rails.

• 03

  Fixed a third-party provider's low success rate

  Average transaction success up 30%; settlement time down 50%.