PWPuck WangSoftware developer

Selected work

Work that balances clarity with velocity.

Full-stack software engineer based in Ottawa, focused on Java/Spring Boot and building reliable systems with Docker, Redis, MySQL, and event-driven architectures.

01 /

Happy Holiday Icon

Cloud-native image generation application that transforms uploaded images into holiday-themed icons through an asynchronous, event-driven AWS processing pipeline.

Stack

Next.jsTypeScriptAWS LambdaAmazon S3Amazon SQSDynamoDBTerraformSharp

Impact

  • Designed an event-driven image processing pipeline with S3, SQS, Lambda, and DynamoDB for asynchronous job orchestration
  • Implemented fault-tolerant processing with conditional state transitions, processing leases, and automated retry recovery
  • Added secure presigned uploads, authentication, rate limiting, prompt-injection safeguards, and CI/CD with Terraform and GitHub Actions
02 /

Wealth Hub Agent

Hierarchical multi-agent system that orchestrates end-to-end financial research, analysis, and decision workflows, with real-time streaming to a web dashboard.

Stack

FastAPIAsyncioNext.jsServer-Sent EventsLlamaIndexChromaDBModel Context Protocol (MCP)

Impact

  • Improved research throughput by 40% via parallel execution of independent agent tasks
  • Standardized tool-calling with MCP for reliable integration with financial data providers
  • Implemented RAG over SEC filings and financial news for semantic search and grounding
03 /

LLM-based YouTube Summarizer

Microservice-based application that generates structured summaries for YouTube videos, including an automated captioning pipeline for videos without subtitles.

Stack

Spring BootSpring AIgRPCAWS LambdaWhisper

Impact

  • Reduced service latency from 500 ms to 100 ms by adopting gRPC
  • Decreased payload size by 30% while improving response performance
  • Built a Whisper-powered Lambda pipeline to generate captions when subtitles are missing