Own the blast radius
Design for failure: idempotency, backpressure, and clear ownership when things break at 2 a.m.
Staff Engineer @ Nykaa · Noida
Backend systems at real scale — auth, ads, logistics.
Ten and a half years shipping distributed platforms: multitenant identity for millions of users, recommendation infrastructure at billions of requests per day, and warehouse automation that moved 100k+ shipments daily. I lead teams, own architecture, and stay close to the code.
10.5+
Years in production systems
1.5B
Daily requests (ads / recommendations)
6
Engineers led (Nykaa)
~30%
Faster auth after platform redesign
How I work
Principles that show up in code reviews and production incidents.
Measure, then optimise
Latency and cost stories beat opinions — profiling, metrics, and targeted fixes over premature cleverness.
Make the path easy
DX and guardrails so the right thing is the default: CI, conventions, and reviews that teach.
Lift the team
Mentoring isn’t extra — it’s how you scale reliability and velocity together.
I’m a Staff Engineer focused on distributed backends, cloud-native architecture, and microservices — from event-driven pipelines to data layers that survive peak traffic. I partner with product and data teams, run hard on system design and reviews, and invest in mentoring so the org ships faster with fewer surprises.
Selected work
Deep dives into problems, approach, and measurable outcomes.
Nykaa · Identity & platform
Multitenant authentication for millions of concurrent users
Context
E-commerce scale identity: security, tenant isolation, and low-latency login paths across the Nykaa ecosystem.
What I did
Architected the next-gen auth & authorisation platform; aligned cross-functional delivery; drove design and code quality; led 6 engineers through SDLC, CI/CD, and security best practices.
Outcomes
- ~30% reduction in authentication time
- ~25% reduction in technical debt (review & refactor focus)
- Stronger security posture for multitenant workloads
Times Internet · Ads & ML serving
Recommendation engine at extreme QPS
Context
High-traffic personalised ad content in real time — latency, memory, and correctness under load.
What I did
Led architecture and development end-to-end; built distributed caching to eliminate data corruption in AdEngine; drove performance work across the stack.
Outcomes
- ~1.5B requests / day sustained
- ~20% uplift in engagement
- ~10GB memory saved; ~40% server efficiency gain
- Zero downtime; consistent data at peak
Swiggy · Supply chain & stores
Warehouse automation & real-time operations
Context
Inwarding, picking, inventory accuracy, and coordination between apps and physical operations.
What I did
Architected inwarding; built Picker App + rack systems; real-time inventory + REST APIs; Control Room rule engine; picker-assignment algorithms; ChitChat / call conferencing.
Outcomes
- ~40% fewer operational errors; 100k+ shipments / day
- ~25% faster order processing
- ~20% fulfillment accuracy improvement
- ~35% faster task allocation
Impact at a glance
Numbers that show up in production — not slide decks.
30%
Faster authentication (Nykaa platform)
1.5B
Requests per day (recommendation / ads)
40%
Server efficiency gain (memory & tuning)
100k+
Shipments / day capacity (warehouse)
Snapshots
Short, true stories from the floor — the ones that still come up in interviews.
Nykaa · Identity
When login became louder than checkout
Traffic grew; auth latency showed up in every funnel review. We redesigned the multitenant path end-to-end — ~30% faster checks — without weakening isolation or audits.
Times Internet · Ads
The cache that stopped bad data at peak
AdEngine saw corruption under load — the worst kind of bug. I led a distributed caching layer so reads stayed consistent; zero downtime through the fix.
Swiggy · Warehouses
From manual chaos to 100k+ boxes a day
Inwarding was still half spreadsheet. We automated QC, receipt, and putaway — ~40% fewer errors and capacity past 100k shipments daily.
Nykaa · Leadership
Six engineers, one bar for quality
As Staff, I didn’t want “my” code to be the only clean code. I paired that with reviews, CI discipline, and ~25% less tech debt where we owned the stack.
Nykaa · Infra
Breaking the monolith DB without breaking trust
User Management lived inside a monolith DB — risky at scale. We moved to a dedicated cluster for HA and shifted hot paths to DynamoDB — resilience and real infra savings.
Swiggy · Ops tech
When the warehouse outran the assignment logic
Pickers were waiting while tasks sat wrongly assigned. A new picker-assignment algorithm cut allocation time by ~35% — felt on the ground the same week.
Career timeline
Every role — most recent first.
Feb 2024 — Present
Staff Engineer
Nykaa · Noida
Multitenant auth platform; team of 6; system design & code quality; HA & DynamoDB migration.
Jul 2022 — Nov 2023
Tech Lead Engineer
Times Internet · Noida
1.5B RPS-class recommendation stack; caching; memory & efficiency wins.
Oct 2018 — May 2022
SDE 2 / SDE 3
Swiggy · Bangalore
Warehouse inwarding, Picker App, inventory, Control Room, picker-assignment, ChitChat.
2013 — 2018
Earlier roles
Medlife — microservices & logistics automation. Rifexions — Finacle customisation for banks. NetCracker — SQL & backend delivery.
Leadership
I mentor for real outcomes: clearer architecture, safer releases, and faster juniors — not just tickets closed.
- Teams — SDLC, reviews, and raising the bar on maintainability.
- Partnership — product, data science, and business — one roadmap.
- Systems thinking — caching, events, and reliability under load.
Stack & architecture
What I ship with — not a buzzword bingo card.
Java
Spring Boot
Go
Python
Kafka
RabbitMQ
MySQL
Redis
DynamoDB
AWS
GCP
Azure
Kubernetes
Docker
Prometheus
Grafana
ELK
Microservices
EDA
CQRS
DDD
Education
B.E., Computer Science
RGPV, Bhopal · 2013
HSC · Science
WBCHSE, Kolkata · 2009
SSC
WBBSE, Kolkata · 2007
Let’s build something reliable.
If you’re hiring for senior backend, platform, or tech lead roles — or want to compare notes on scale — I’d like to hear from you.