DEVOPS WORK & CASE STUDIES
Detailed technical breakdowns of cloud infrastructure, serverless architectures, CI/CD automation pipelines, and software systems.
Serverless Online Voting Platform
The college relied on paper-based physical voting for over 500+ students, resulting in manual vote tallying errors, slow result delivery, and vulnerability to tampering.
DevOps Engineer & Backend Developer — designed the AWS serverless architecture, Terraform IaC scripts, and GitHub Actions automated deployment pipeline.
- ❯Wrote 6 AWS Lambda REST microservices using Python.
- ❯Configured API Gateway routes with IAM least-privilege policies.
- ❯Provisioned DynamoDB On-Demand for automatic peak traffic scaling.
- ❯Built GitHub Actions CI/CD pipeline reducing deploys from 8 manual steps to zero.
- ✓500+ active students served with zero downtime during peak voting hours.
- ✓180ms P99 latency recorded across all REST API endpoints.
- ✓<0.1% error rate during live operations.
- ✓Reduced deployment execution time from 12 minutes to 3 minutes.
Dorm-Dish Multi-Tier AWS Platform
Traditional EC2 server hosting incurred continuous monthly costs even during idle periods, while lacking automated scaling for high campus demand spikes.
Cloud Architect — designed CloudFormation stacks, API Gateway integration, and CloudFront global CDN distribution.
- ❯Refactored monolithic backend into serverless microservices.
- ❯Configured S3 static website hosting fronted by CloudFront edge caching.
- ❯Wrote CloudFormation IaC templates for reproducible environment spins.
- ✓Achieved an 80% cost reduction compared to provisioned EC2 instances.
- ✓100% automated infrastructure setup via CloudFormation.
- ✓Sub-second global content delivery via CloudFront edge locations.
Statement Dashboard PWA
Users hesitated to upload sensitive PDF/CSV bank statements to cloud servers due to privacy concerns and third-party data collection risks.
Full-Stack Architect — engineered client-side PDF parsing using pdf.js, local WebStorage persistence, and offline PWA service workers.
- ❯Integrated React 19 + TypeScript + Vite for client-side execution.
- ❯Built offline PDF/CSV parsing pipeline with zero server transmission.
- ❯Implemented recurring payment pattern detection algorithms in browser.
- ✓100% offline security — zero byte data leaves the client device.
- ✓Sub-500ms statement parsing speed for 50+ page PDFs.
- ✓Complete financial health scoring engine running in local Web Worker.
AI Resume-Job Matcher
Job applicants faced high rejection rates due to keyword mismatches and unoptimized resume formatting against Automated Tracking Systems (ATS).
ML & Systems Engineer — trained semantic embeddings model on 1,200+ tech job descriptions and deployed Streamlit app.
- ❯Extracted TF-IDF and sentence-transformer semantic vector embeddings.
- ❯Trained Logistic Regression classifier achieving high AUC-ROC metric.
- ❯Built Streamlit interactive UI with instant keyword gap analysis.
- ✓90% classification accuracy on test dataset of tech job postings.
- ✓95% AUC-ROC score for candidate-job relevance matching.
- ✓Instant actionable ATS keyword suggestions for applicants.