AWS cost anomaly triage
Turned SNS/SQS cost alerts into investigations that collect Cost Explorer, CloudWatch, CloudTrail, tags, and CUR evidence.
- Output
- structured incident reports
- Added
- S3 cost visibility through Athena
Pranav Arora / DevOps Engineer
I keep production systems boring.
Kubernetes, AWS, CI/CD, service mesh debugging, and the cleanup work that keeps incidents from becoming drama.
I am a DevOps Engineer at Unicommerce. Most of my work sits close to production: migrating services to Kubernetes, reducing noisy AWS cost alerts into evidence, tuning traffic behavior, and making database operations safer.
10,000+ sellers
20+ app servers to K8s
30% faster build pipeline
field notes
real work
selected cases
Istio traffic concentration
Investigated why least-request balancing favored newly scaled pods, then worked toward steadier request distribution.
- Signal
- traffic packed onto new pods
- Result
- routing behavior stabilized
Kubernetes migration
Automated migration of 20+ application servers and added autoscaling paths for variable production load.
- Scope
- 20+ application servers
- Added
- HPA and VPA coverage
where I spend time
- Kubernetes
- migration, autoscaling
- AWS
- cost, edge, storage
- Data ops
- MySQL, S3, Athena
how I approach issues
- 01 start from logs, metrics, and traffic shape
- 02 write down the hypothesis before changing systems
- 03 sequence risky work so rollback stays possible
Numbers from the resume
The short version.
Selected work
Work I can explain without hand-waving.
A few pieces from production DevOps at a high-scale e-commerce SaaS platform serving 10,000+ sellers.
Tools I use
The stack is practical, not decorative.
Infrastructure first, with enough automation and AI tooling to remove repeated manual investigation work.
Timeline
Where the work happened.
Contact
Let's talk about what you're building.
Have an idea, opportunity, or infrastructure challenge in mind? Feel free to reach out; always open to meaningful conversations and collaboration.