Sector 02 · Path
Netradyne, Bengaluru. I started with 150+ services and a monitoring setup that alerted on everything and explained nothing. Now I'm the primary on-call owner for 215+.
The highway
Feb 2025 → presentFeb 2025
Cloud Intern
Netradyne, Bengaluru
- Inherited 150+ services and a monitoring setup that alerted on everything and explained nothing.
- Built AlertFlow in month 2 and Exception Clustering in month 4.
Mid 2025
Associate Site Reliability Engineer
Netradyne, Bengaluru
- Converted at month 6.
- Built TraceLens: triage from about 2 hours to about 15 seconds.
Projects from this role:TraceLensFeb 2026 – present
Software Engineer, SRE & AI Tooling
Netradyne, Bengaluru
- Primary on-call owner for 215+ AWS microservices. Led Sev-1 and Sev-2 bridges and prevented 3 Sev-1/P0 incidents by correlating early signals across logs, metrics and traces.
- Built TraceLens, LogAgent, AlertFlow and Exception Clustering, plus latency and traffic anomaly detection and SLO/SLI automation that alerts before burn rate turns critical.
- Sole engineer on the PostgreSQL to DynamoDB migration of critical production data, with a zero-downtime cutover. Core architect of organisation-wide multi-cluster EKS enablement.
- Owns the observability stack (Prometheus, Grafana, ELK, CloudWatch), n8n automation and GitHub production-diff tracking on EKS. Cut tooling spend by replacing paid SaaS with internal builds.
How I work
- 01
Boring is a feature.
The best incident is the one that never pages anyone.
- 02
Delete more than you add.
Every line of code is a liability at 3 AM.
- 03
Automate the pain, not the fun.
Do a toil task twice and the third time is a script.
- 04
Show the work.
An agent that can't explain its reasoning is a liability; every TraceLens run is auditable step by step.
Stack constellation
hover a starHover a star
Each star is a tool I use. Hover or focus one to light the projects and roles that use it, here and across the site.