Hello.

My name is Leonard - a software engineer who builds and ships scalable, event-driven systems from backend services to cloud infrastructure. I work across Node.js, TypeScript, Python, and Go, and specialize in distributed systems, real-time processing, and cloud-native deployments on AWS - with a strong focus on observability, reliability, and long-term maintainability.

Connect with me on LinkedIn.



Proactive Attitude

Takes initiative and doesn’t wait to be told what to do – shows ownership and drive.



Reliable Communication

Keeping you updated at every step to ensure transparency and clarity.



Problem-Solving Mindset

Turning obstacles into opportunities with strategic and creative thinking.

Languages / Technologies



Languages

JavaScript
TypeScript
Python
Go



Frontend

React
Next.js
Tailwind CSS
HTML/CSS
Shadcn



Backend

Node.js
Bun
ElysiaJS
Express.js
Koa.js
Hono
PostgreSQL
NoSQL
GraphQL



DevOps

AWS (ECS Fargate, CloudFront, EC2, Lambda, S3, IAM, etc.)
CD/CD (GitHub Actions)
Docker
Kubernetes
Terraform
Linux
Helm
FluxCD
OpenTelemetry

Projects

Content Repurposing Engine

Content Repurposing Engine is an event-driven AI video pipeline designed to automatically transcribe, analyze, and render optimized short-form clips from raw media. It is built as a cloud-native monorepo using a Bun API, Python rendering workers, and Terraform on AWS.

Architected an event-driven video processing engine, replacing persistent background polling queues with ephemeral AWS ECS Fargate tasks triggered via ecs:RunTask, optimizing idle cloud compute costs by over 40%.
Provisioned complete infrastructure as code using Terraform, deploying VPCs, ALBs, S3, and CloudFront with strict security groups to support a highly available, decoupled monorepo architecture.
Implemented a secure CI/CD pipeline using GitHub Actions and AWS OIDC, automating Python/TypeScript linting, multi-stage Docker builds, infrastructure drift detection, and ECR image deployment – eliminating long-lived AWS credentials.
Orchestrated a cost-optimized AI pipeline integrating Claude, Gemini, and GPT-4o for intelligent content analysis, Groq (Whisper) for high-speed transcription, and FFmpeg for short-form clip rendering.

Tech Stack: TypeScript + Python + AWS ECS Fargate + Terraform + GenAI + CI/CD
Repository: GitHub

Distributed Webhook Dispatcher

A fault-tolerant webhook dispatcher built in Go, designed to sustain 10,000+ RPS with sub-100ms p99 latency. It features Goroutine-based worker pools, circuit breakers, and a PostgreSQL-backed dead-letter queue deployed to Kubernetes via GitOps.

Designed and built a fault-tolerant Go webhook dispatcher sustaining 10,000+ RPS with sub-100ms p99 latency using Goroutine-based worker pools.
Built retry orchestration with exponential backoff, circuit breakers, and PostgreSQL-backed dead-letter queue to guarantee zero message loss.
Containerized using multi-stage Docker builds and deployed to Kubernetes via Helm and FluxCD (GitOps).
Instrumented with OpenTelemetry for metrics and tracing to monitor latency, retries, and failure rates.

Tech Stack: Go + Concurrency + Kubernetes + Observability
Repository: GitHub

Serverless URL Shortener

A serverless URL shortening service built on AWS Lambda, API Gateway, and DynamoDB. It supports TTL-based link expiration, click analytics, and rate limiting — fully provisioned with SST and automated via GitHub Actions.

Architected a serverless URL shortening service using AWS Lambda, API Gateway, and DynamoDB.
Implemented TTL-based link expiration, click analytics tracking, and rate limiting for abuse protection.
Provisioned infrastructure using SST (Infrastructure as Code) and automated deployments via GitHub Actions.
Optimized for cost efficiency under AWS Free Tier with event-driven execution.

Tech Stack: AWS + IaC + CI/CD
Repository: GitHub