Frontend Engineer · carrier-grade telecom

LotusFlare - DNO Cloud Portal

I own the UIs that let business analysts configure complex data pipelines, reporting and real-time event processing for carrier-grade telecom operators serving millions of subscribers. The work is about making genuinely hard configuration feel manageable - and keeping a large Vue codebase fast and consistent.

80%

faster initial load (1.2 min → 14 s)

45%

less complexity, zero regressions

500+

files brought to one standard

draw.io SVGs → interactive Vue Flow diagrams

Context

The DNO Cloud Portal is how operators serving millions of subscribers configure data pipelines, reporting and real-time event processing. LotusFlare is Ericsson-backed, and the stakes match: the interfaces are dense and analysts drive carrier-grade systems through them. My remit is the frontend that makes that power usable.

Performance: 1.2 minutes → 14 seconds

Initial load had crept to roughly 1.2 minutes - unacceptable for a tool people live in all day. I cut it by ~80% to about 14 seconds through dynamic imports, route-level code splitting and Vite build optimizations, shrinking what the browser has to parse and execute before the portal is usable.

Modernizing the codebase

Led the Vue 3 Composition API + Pinia migration and authored reusable composables adopted across the team.
Refactored a 2,000-line SinkConfigEditor into a modular, step-based flow - 45% less complexity, zero regressions.
Established codebase standards and reorganized imports across 500+ files for long-term maintainability.
Led the Dataflows migration across frontend and Scala backend: five draw.io SVGs became interactive Vue Flow diagrams as backend-owned content, behind a per-operator flag with a contract test and an ADR. Legacy path deleted.
Upgraded the toolchain to Vite 8 and Vitest 4, and added an oxlint + ESLint pipeline that cut lint CI time.
Architected the team's AI engineering stack: cross-IDE agent rules, a CLAUDE.md context system and custom skills.

Features that ship real capability

A multi-file S3 upload system with real-time progress, smart error handling and DLQ recovery.
Role-based access control for Reports, with permission-driven UI rendering.
An Application Management UI with cluster-level actions for data-pipeline orchestration.

The throughline

The 80% load-time cut and the zero-regression refactor are the measurable edges of the work, and the AI engineering stack I set up is how the whole team moves faster on it.