Atul Bhardwaj

About Me

CS grad, cloud builder, two apps live in production.

I'm finishing my CS master's at Seattle University and spending most of my time building on AWS. I have two apps running in production right now — a serverless RAG chatbot and a three-tier platform — both built to solve real problems for real users on campus.

I lead the AWS Cloud Club at SU, where I grew the community from 13 to 90+ members, secured $10K in AWS credits, and co-hosted a 24-hour hackathon with UW after cold-emailing Jeff Barr. I like figuring out the right architecture for a problem — not just the most popular one.

I'm 3x AWS certified and a BeSA Agentic AI graduate (Strands Agents SDK, Bedrock AgentCore, MCP). I'm on STEM OPT through mid-2029.

AWS Certified

3.6+

GPA

Live AWS Apps

90+

Cloud Club Members

Professional Experience

Shipping production AWS applications and leading teams

Cloud Solutions Lead

AWS Cloud Club, Seattle University

Jan 2026 – Present · Seattle, WA

Scaled the club from 13 to 90+ members in two quarters by building partnerships with AWS Solutions Architects, Developer Advocates, and the AWS Skills Center, turning cold outreach into recurring collaboration.
Built a direct relationship with AWS VP Jeff Barr through cold outreach, securing $10K in AWS credits and a co-hosting partnership with UW for a 24-hour, three-track hackathon (Build with Bedrock, Serverless Solutions, Cloud for Good).
Evangelized AWS services (Bedrock Knowledge Bases, Lambda, API Gateway) through cloud architecture workshops for 20+ builders per session, helping participants select scalable, cost-effective solutions for their own workloads.

Data Science Consultant

Statistics Without Borders

Sep 2025 – Apr 2026

Served as technical point of contact across 7 teams (14 contributors, 14+ datasets), distilled data quality issues into clear technical requirements, and presented standards to non-technical leadership that the organization adopted.
Migrated 14 contributors from unversioned email workflows to GitHub, accelerating adoption by authoring onboarding docs that got non-technical members committing in under a week.

Operations Technician

Seattle University

Oct 2024 – Present · Seattle, WA

Gathered stakeholder feedback and cross-referenced it against parking and transit utilization data, identified 7 plans adding cost without serving demand, and translated findings into a consolidation recommendation leadership approved.
Found a manual bottleneck in ID card fulfillment (staff hand-sorting 200+ cards per batch), rewrote the Entrust batch SQL to auto-sort at query time, eliminating the manual step entirely.

Featured Projects

Solving real-world problems with data and code

AWS Cloud — Live Production

SUChatBot — Campus RAG Chatbot

2025 · Live

Replaced 53 static SharePoint pages with a serverless chatbot that answers questions for 5 campus departments instantly — no staff needed.

S3 Vectors over OpenSearch — ~$350/month saved, sub-3s responses
Fully serverless: Lambda + API Gateway + CloudFront
Claude Haiku 4.5 for inference; Titan Embeddings V2 for vectors

Bedrock Lambda S3 Vectors API Gateway CloudFront RAG

Live App →

Lost & Found Platform

2025 · Live · $0.90/day

Three-tier AWS app connecting 4 campus offices under one lost-and-found system. Chose EC2/RDS over serverless to keep it always-on and snappy.

Custom VPC with public/private subnets across two AZs
Cognito auth locked to @seattleu.edu; IAM least-privilege throughout
CI/CD via GitHub Actions; nginx + Flask + React + PostgreSQL

EC2 RDS VPC Cognito CloudFront IAM GitHub Actions

Live App →

AI / ML

NeatCode — AI Code Refactoring

March 2026

Paste messy code, get it cleaned up by Gemini AI with an explanation for every change — so you actually understand what changed.

Dockerized Flask on Elastic Beanstalk; React on S3 + CloudFront
Input guard rails reject non-code submissions
Built with a team of 3 (Agile), then extended and deployed solo

Flask React Gemini AI Elastic Beanstalk Docker CloudFront

GitHub →

Amazon Lex Banking Chatbot

November 2025

Conversational chatbot for banking transactions — custom intents, slot types, and multi-turn dialogue, all wired to Lambda for live responses.

Lambda integration for dynamic, context-aware responses
Deployed via CloudFormation (infrastructure as code)

Amazon Lex Lambda CloudFormation

Data Engineering & Cloud

YouTube Trending Analytics AWS Data Lake

YouTube Trending Analytics — Data Lake

November 2025

Serverless ETL pipeline that ingests 1,000+ trending videos across 5 regions and makes them queryable in seconds.

Three-tier S3 lake: Raw → Cleansed → Analytics
75% storage reduction · 80% query cost savings via Parquet
Athena SQL + QuickSight dashboards

S3 Glue Athena PySpark Parquet

GitHub →

Real-Time TF-IDF Search Engine

March 2025

Distributed search engine that ranks large-scale text in real time using TF-IDF, built on Spark across an AWS multi-service stack.

Apache Spark on EMR for parallel processing
EC2 · S3 · DynamoDB · Lambda backing the query layer

Apache Spark EMR DynamoDB Python

Data Science & Analytics

Campus Operations — Behavioral Segmentation

September 2025

Clustered campus department behavior patterns to surface where operations could be trimmed without hurting service quality.

5 distinct segments via GMM + K-Means
25% reduction in unnecessary support tickets

Python K-Means GMM Scikit-learn

Technical Skills

Tools and technologies I work with

Architecture

Multi-tier Design Serverless Patterns VPC Networking Cost Optimization High Availability CI/CD Infrastructure as Code Migration

AWS Services

EC2 S3 RDS VPC Lambda CloudFront Cognito DynamoDB Bedrock API Gateway S3 Vectors IAM Glue Athena EMR CloudFormation Elastic Beanstalk

AI & ML

Bedrock Knowledge Bases Bedrock AgentCore Strands Agents SDK RAG Titan Embeddings V2 FAISS MCP Scikit-learn TensorFlow PyTorch

Data & Analytics

PostgreSQL MySQL DynamoDB Apache Spark PySpark ETL Pipelines Data Modeling Power BI Tableau

DevOps & Programming

Python JavaScript SQL Bash Docker Git GitHub Actions nginx Linux

Research Publications

Peer-reviewed contributions to the field

Evaluating ML Algorithms for Heart Disease Prediction

IEEE Xplore 2024 Read Paper →

Comprehensive evaluation of six supervised ML algorithms for heart disease prediction, optimizing precision, recall, and F1-score to minimize false negatives in clinical decision support systems.

Bidirectional LSTM for Toxic Comment Classification

EAI Journal 2024 Read Paper →

Novel Bidirectional LSTM + Convolution approach for identifying and classifying toxic comments online, demonstrating improved NLP performance across benchmark datasets.