The Engineer
Five years. Billions of events.No dropped records.
Data correctness is the product; not a post-launch concern.
Why I build, and what taught me to build it right.
I got into engineering to build things people actually need. At D.Y. Patil I got to do that again and again: a coordination app for a national-level event our college hosted, and an app that gave villagers one place to reach essential services.
At TCS, honestly, I made mistakes, and every one taught me how systems behave at scale: one bad load can quietly break every report downstream. That's where I understood data engineering isn't just a job. It's understanding how the whole system works, because a lot of people depend on it.
I went to USC to go deeper, into distributed systems, machine learning and where they meet. At Visa I owned several pipelines end to end, from ingestion to the features ML models trained on, and learned that the hard part isn't moving data fast, it's being able to rerun it safely. I built them to be idempotent and made sure data quality was never compromised. At Citi, I build Snowflake platforms and an AI agent that investigates data issues on its own.
All of that changed how I think. I don't just ask how to fix a problem today; I ask how the fix holds up a year from now, when the data has doubled and the people relying on it have changed. I use AI every day to move faster, but nothing ships until I've reviewed it, because if people are going to trust the data, I have to trust the code first.
The Pipeline
From CS fundamentals to real-timepipelines — and now AI agents.

Data Engineer · AML
Running automated Python, PySpark and Snowflake pipelines that process 8M+ records a day for AML analytics. Built a Snowflake Cortex AI agent that investigates data-quality anomalies on its own, cutting manual triage 30%; moved 25 on-prem pipelines to Snowflake with AI-assisted conversion (−40% effort); and trimmed Snowflake credit spend 18% without missing an SLA.
Data Engineer · Payments & Fraud
Built idempotent Delta Lake pipelines on Azure Databricks for 10M+ card transactions a day from 8+ payment sources, plus Kafka-to-Delta streaming that feeds near-real-time fraud scoring. Moved legacy Hive jobs to PySpark (4 h → 90 min), published leak-free feature tables for fraud models, and added quality gates that cut data incidents 35%.

M.S. Computer Science
Advanced study in software systems and machine learning — efficient algorithms, scalable architectures, and ML techniques for complex, data-driven problems. Where data engineering and AI merged for me.

Data Engineer
Built PySpark ETL on AWS EMR joining batch claims, provider and member data with Kinesis-streamed eligibility events — 60M+ records a month into an encrypted S3 lake. Tuned Redshift for 28% faster dashboards, modeled SCD2 star schemas, and used Neo4j graphs to surface potential healthcare collusion.

B.E. Computer Science & Engineering
Built a strong foundation in CS fundamentals, with a focus on software development and systems engineering.
Things I build
Data platforms I design and buildend to end, on public or simulated data.
Recognition
Awards, leadership rolesand a few things off-résumé.
My Expertise
Stream-first, cloud-agnostic,warehouse-native.
Certifications
Verified credentials &completed coursework.

Writings
Hard-won lessons frompipelines under pressure.


