About

The resume, with the actual person still intact.

I moved from IIT Kharagpur to New York, spent years in enterprise and Amazon-scale systems, and then stepped into a builder phase where product conviction and ownership matter more than looking conventionally impressive.

Three-act arc

India, New York, then the builder phase.

India · 2013–2021

Foundation

IIT Kharagpur Electrical Engineering, KVPY scholar. Then Loylty Rewardz in Mumbai, leading a batch→Kafka migration on a loyalty engine processing millions of transactions a day. First real production chops.

NYC · 2021–2025

Enterprise

NYU Tandon MS. Then Amazon Ads — Java microservices powering international marketplace expansion and India-specific ad placements, migrating services to native AWS across its CI/CD ecosystem. Distributed-systems discipline at scale.

Now · Sep 2025 → present

Builder

Building software for businesses full-time. Two live client sites, two AI products, and selective on what's next.

Start from problems, not tech stacks. Don't just write code — ship products.

Resume structure

The formal story and the current one are finally aligned.

Experience

The work history, chronologically.

Amazon-scale systems, enterprise backend foundations, and then a builder phase where product taste started to matter as much as technical depth.

2022–2025 · New York, NY

Software Development Engineer · Advertising

Amazon

Java microservices for ad delivery and campaign management; owned new ad placements for Indian marketplaces as part of international expansion.
Streamlined new-marketplace placement launches — added new services and centralized cross-package configuration (e.g., AWS AppConfig) — cutting the end-to-end launch timeline from ~4 months to ~1 month.
Migrated advertising services from MAWS to NAWS (native AWS), working across the AWS ecosystem and its CI/CD tooling — ECS, CDK, deployment pipelines, and related services.

2017–2020 · Mumbai

Software Development Engineer

Loylty Rewardz

Owned backend services across the full transaction lifecycle — multi-source ingestion (batch files and real-time vendor integrations), rule-engine processing, point awarding, OTP-verified redemption, and event- and time-based campaign services, with UI integration where needed.
Automated and optimized partner file processing across all banks and retailers using Talend ETL and Apache Camel tuned in Java, with cron-scheduled shell scripting and reconciliation review queues that eliminated record loss.
Built a monitoring service and portal integration for real-time health, stability, and availability across servers, resources, and APIs spanning all microservices.

Education

IIT Kharagpur and NYU Tandon are both in the wiring.

NYU Tandon School of Engineering

MS, Electrical Engineering

ML, Deep Learning, Big Data coursework.

IIT Kharagpur

BTech, Electrical Engineering

Computer Vision, IoT projects. KVPY scholar.

Working mode

The way I work now.

I like end-to-end ownership, products with a real point of view, and systems where the operational details are part of the product rather than hidden behind it.

Skills

The stack is wide because the work has been end to end.

Languages

JavaPythonJavaScriptTypeScriptSQL

Backend

SpringHibernateApache KafkaApache CamelTalend (ETL)rule-engine / event-drivenNode / Express

Frontend

ReactNext.jsvanilla JS when it's the right call

AI / ML

LangChain patternsReAct loopsMCPMulti-provider routingSentence-transformerspgvector retrieval

Data

Parquet

Ops

AWS ecosystem & CI/CD (ECS, CDK, AppConfig)NginxLet's Encryptmonitoring & alertingDeploy pipelines you can hand off

Builder bias

The throughline is not “full stack.”

It is owning enough of the product to make better decisions: backend systems, AI orchestration, interface quality, deployment, and the narrative layer people actually experience.