Admin &
Support Portal
Two portals at Evren — the Admin Portal (customer-facing interface for enterprise OS management, rebuilt from scratch as V2) and the internal Support Portal used by engineers to manage enterprise customer environments. Migrated backend from Python to Node.js, designed 4-tier RBAC, overhauled the React UI, and built a log-analysis tool that cut debugging time by 50%.
A Python backend that couldn't scale, no meaningful access controls, and debugging that required tailing raw logs manually.
Evren's Admin Portal is the primary customer-facing interface for enterprise OS management — clients use it to configure, monitor, and control their operating environments. The original V1 ran on a Python backend that couldn't scale: response times degraded under load, onboarding engineers to the codebase was slow, and access control was a flat admin flag — every admin had identical permissions regardless of role or scope.
On top of that, debugging production issues meant manually grepping through server logs. There was no structured log analysis tooling, no visibility into error patterns, and no way to correlate API latency with specific operations — each incident was a blank-slate investigation.
Node.js migration + RBAC redesign + React UI rebuild — shipped without downtime.
Migrated the backend and redesigned access controls without touching production uptime.
Rather than a big-bang rewrite, the migration ran the Python and Node.js APIs in parallel behind a routing layer. Routes were migrated one service at a time, verified in staging, then cut over. By the time Python was fully decommissioned, Node.js had been running in production for months. API latency dropped 30% — partly from Node's async I/O model, partly from query optimisations that the migration surfaced.
The old system had one admin boolean. The redesign introduced four tiers — Super Admin, Team Admin, Support Agent, and Viewer — each with a distinct permission set enforced at the API layer via a policy engine, not just on the frontend. Super Admins manage the full enterprise scope; Team Admins control their org's systems; Support Agents action tickets without access to infrastructure settings; Viewers have read-only access. Portal adoption jumped 40% after launch, largely because staff could now do their jobs without escalating every second action.
Built a Python tool that ingested raw application logs, parsed them into structured events (request, error, latency, user context), and produced daily summaries plus live alerts for anomalies. What previously required 30 minutes of grep and log correlation became a 2-minute dashboard query. Debugging time dropped 50% and on-call engineers stopped dreading production incidents.