Who is Soumya Swaraj?

Soumya Swaraj is an AI & Cloud Architect and founder of Evergrow. He is a former JPMorgan Chase engineer and holds six AWS certifications. He specializes in engineering scalable data platforms, distributed systems, and agentic AI workflows.

Evergrow is an AI Content OS founded by Soumya Swaraj. It uses LLM agents, RAG pipelines, and scalable cloud infrastructure to automate content operations end-to-end.

What did Soumya Swaraj build at JPMorgan Chase?

At JPMorgan Chase, Soumya architected enterprise-scale data platforms handling 10M+ daily requests on a stack of Aurora Serverless, Amazon Kinesis, AWS Lambda, EC2, and Firehose. He was an early adopter of LLM agents for anomaly detection and self-healing systems, and built on-demand natural-language-to-SQL engines that compressed 10 days of analyst work into 10 minutes.

How many AWS certifications does Soumya Swaraj hold?

Soumya Swaraj holds six active AWS certifications covering solutions architecture, data, and specialty domains.

What is Soumya Swaraj's track record?

His shipped work includes enterprise data platforms for JPMorgan Chase, Early LLM agents for monitoring feeds, anomaly detection, advanced reporting, automated workflows and self-healing systems, Content OS for Evergrow; Earlier school & college time projects include a movie recommendation site that drew 15M hits, a RPG gaming community that reached 15,000 members — and a COVID-19 tracker that served 800,000+ users during the pandemic. He also built university-scale platforms, CMS for Virginia Tech and Thapar Institute, and multiple campus applications.

How can I hire or contact Soumya Swaraj?

Soumya accepts a limited number of engagements per year. You can book a call directly via the in-site scheduler, or reach him on LinkedIn at linkedin.com/in/isoumyaswaraj.

What technologies does Soumya Swaraj specialize in?

He specializes in AWS (Aurora Serverless, Kinesis, Lambda, EC2, Firehose), agentic AI workflows, LLM integration, RAG pipelines, distributed systems, scalable data platforms, and natural-language-to-SQL engines.

Soumya Swaraj — AI & Cloud Architect

Soumya Swaraj — AI & Cloud Architect I build AI that works at scale.

I'm Soumya Swaraj — founder of Evergrow and a former JPMorgan Chase engineer. I help companies take AI from prototype to production: LLM agents, RAG and agentic workflows on AWS that stay fast, reliable and affordable as usage grows.

Hi, I'm Soumya.

I design AI and cloud systems for companies that can't afford for them to break.

I learned the craft at JPMorgan Chase, building data platforms inside a bank that moves billions of dollars a day. Today I run Evergrow, where teams of AI agents research, write and fact-check content made for AI search.

Based in

Bengaluru, India

Now

Founder, Evergrow

Before

JPMorgan Chase

Clients

Worldwide

Available for select AI & cloud engagements in 2026.

Zero downtime, by design.

Decoupled, event-driven AWS platforms built to absorb traffic spikes and survive the loss of a zone without dropping a request. Run either scenario below.

Pick a scenario to watch the platform respond.

Scale isn't something you add later.

Back in high school, I built a gaming community that grew to 15,000 members, then watched it collapse on shared hosting. The code was fine. The system around it wasn't.

Every project since has started with one question: what happens when this works? More users, more data, more agents, more money moving through it. I design for that day first, so growth shows up as revenue instead of an outage.

Principle 01

Design for the day it works.

Capacity, cost and failure modes get planned before the first user arrives, not after the first outage.

Principle 02

Make failure boring.

Good systems detect, diagnose and fix the common problems themselves. People handle the rare ones.

Principle 03

Measure what changed.

Revenue earned, hours saved, tickets closed. If it can't be measured, it isn't finished.

EvergrowFounder2025 — now Live

Making brands the answer AI gives.

$200K revenue in 60 days for a B2B SaaS client

7× organic traffic in 100 days for tourism agencies

The problem: Search is moving from ten blue links to one AI answer. Most brands aren't in that answer, and generic AI content won't put them there.
What I built: An AI content OS: a multi-agent system where adversarial AI personas research, challenge and fact-check every piece before it ships. We run it as a done-for-you service for startups and established brands.
The result: Content that ranks on Google and is written to be recommended by ChatGPT, Gemini and Perplexity, with revenue to show for it.

Written for OpenAI

Multi-agent
Adversarial review
RAG
GEO

Visit Evergrow

JPMorgan ChaseSoftware Engineer2022 — 2024 Shipped

Freeing a bank's reports from its most expensive hardware.

10M+ requests routed every day, with zero downtime

$1M+ in yearly savings the platform work contributed to

The problem: Thousands of accounting, settlement, regulatory and risk reports lived on in-memory clusters built for real-time trading. Fast, but costly, and some datasets ran to terabytes.
What I built: A decoupled, cloud-native replica on AWS that took over every workload that didn't need microsecond latency. I designed the streaming pipelines, the consumer SDK and the migration tooling, plus anomaly detection and self-healing.
The result: 10M+ requests a day with zero downtime, in an environment that moves billions of dollars daily.

Aurora Serverless
Kinesis
Firehose
Lambda
S3
SNS
Athena

JPMorgan ChaseAgentic LLMs2022 — 2024 Shipped

AI agents inside a bank, before most banks tried them.

10 min to answer data questions that used to take analysts 10 days

100M+ trade requests a day on the platforms the agents watched

The problem: Data anomalies in financial reporting were caught late, often by the teams downstream who relied on the numbers. And every new data question cost days of analyst time.
What I built: LLM monitoring agents that detect, diagnose and resolve anomalies on their own. Plus on-demand natural-language-to-SQL agents, grounded with RAG and few-shot prompting.
The result: Anomalies resolved before they reached anyone downstream. Ten-day analyst requests answered in ten minutes.

LLM agents
RAG
Few-shot prompting
Text-to-SQL
Self-healing

JPMorgan ChaseAutomation Shipped

400 batch feeds, tuned until they fixed themselves.

1.5M+ compute hours saved

80% fewer daily production tickets

The problem: More than 400 batch feeds carried millions of transactions a day across fixed income, commodities, derivatives and regulatory reporting. Every failure became someone's morning ticket.
What I built: End-to-end optimization of the feeds, then automated self-correction so the common failures repair themselves.
The result: 1.5M+ compute hours saved, and 80% of daily production tickets gone.

Batch processing
Self-correction
Cost optimization

Amazon Web ServicesCertifications Verified

Six AWS certifications, from networking to machine learning.

Issued by AWS across architecture, networking, machine learning and development. Open any certificate to view it.

Solutions Architect Professional

Advanced Networking Specialty

Machine Learning Engineer Associate

Solutions Architect Associate

Developer Associate

Cloud Practitioner

Beyond the
architecture.

Building scalable systems requires intense focus — but the best ideas often happen away from the screen.

When I'm not architecting AI ecosystems or optimizing cloud infrastructure, I'm usually resetting my competitive drive on the badminton court or diving into a good fiction novel. I prefer "chill" travel — finding quiet hill stations or beaches over high-intensity trekking — and I am perpetually on the hunt for the perfect cup of Adrak Chai.

Badminton Fiction Chill Travel Adrak Chai

Soumya Swaraj — AI & Cloud Architect: I build AI that works at scale.

Hi, I'm Soumya.

Zero downtime, by design.