Case Studies
Projects I've shipped
Real projects, real outcomes. Each of these represents a problem solved, a system shipped, and a team that moved faster after we worked together.
Serverless Web Application Architecture
Problem: A growing SaaS startup needed to replace their monolithic backend with a scalable, cost-efficient serverless architecture serving 50k+ daily users.
Approach: Designed a fully serverless stack using API Gateway, Lambda, DynamoDB with single-table design, and CloudFront for edge delivery. Implemented infrastructure as code with AWS CDK and automated CI/CD pipelines with canary deployments.
Outcome: Reduced infrastructure costs by 72%, improved p99 latency from 800ms to 120ms, and eliminated all after-hours ops pages. The team went from monthly deploys to deploying multiple times per day.
Cloud Deployment Automation
Problem: An established business had a manual deployment process involving SSH, bash scripts, and late-night maintenance windows. Deployments were error-prone and took hours.
Approach: Containerized their application, set up automated CI/CD pipelines with GitHub Actions, and deployed to AWS ECS Fargate with rolling updates. Added automated smoke tests and CloudWatch dashboards for observability.
Outcome: Deployment time dropped from 3+ hours to under 10 minutes with zero-downtime rollouts. Rollbacks became a single click. The team regained their evenings and weekends.
AI-Assisted Internal Tooling
Problem: A logistics company's operations team was spending 15+ hours per week manually processing shipment data across spreadsheets, emails, and a legacy ERP.
Approach: Built a lightweight internal tool with a natural language interface powered by an LLM. The system ingests data from multiple sources, normalizes it, and lets the operations team query and update it conversationally.
Outcome: Reduced manual data processing from 15 hours/week to 2 hours/week. Operations team adoption was immediate because the interface matched how they already thought about their work.
Database-Backed Business Workflow App
Problem: A professional services firm relied on a patchwork of spreadsheets and email to track client engagements, deliverables, and billing. Data was inconsistent and reporting was painful.
Approach: Built a custom web application with a relational data model, role-based access, automated notifications, and a clean dashboard. Deployed on AWS with a serverless backend for low maintenance overhead.
Outcome: Centralized all client engagement data into a single source of truth. Reporting that used to take a full day now happens in real time. The firm scaled 2x without adding administrative overhead.
Have a project in mind?
Every project starts with a conversation. Tell me what you're working on and I'll let you know if I can help.