Duan Wenbo Alfred

Software Engineer Intern – AI Agent

• Designed and shipped an AI Chat Agent for enterprise customers, owning the AI workflow and backend from prototype to production
• Built the FastAPI service behind the agent and ran it on Docker and Kubernetes with autoscaling and health checks, keeping p95 latency under 500 ms for more than 10K requests each month
• Shaped the RAG and model stack with LangChain, FAISS, LoRA and TensorRT and wired it into GitHub Actions, Prometheus, Grafana and OpenTelemetry

Software Engineer Co-op Project

• Improved live music and audio for users on weak networks by tuning a multi-threaded C++ and Swift pipeline, cutting end to end streaming delay and making playback feel smoother
• Added a real time orchestration layer with Kafka, Redis and WebSockets that could fan out live events across services
• Shipped an AI powered content discovery feature using PyTorch and FAISS so users could reach the tracks they liked faster, which lifted daily active users and reduced navigation time

Software Engineer Intern

• Built a distributed web crawler in Node.js using Express, TypeScript and RabbitMQ that could process more than thousands pages per minute, deployed on AWS EC2 with Docker Swarm and Jenkin
• Designed MongoDB and PostgreSQL schemas for terabyte scale data and tuned indexes and queries so most lookups returned in a few milliseconds and stayed consistent under load
• Shipped a React dashboard where customers could watch crawler jobs and system health in real time

Software Engineer Intern

• Exposed REST APIs that let field engineers pull device data on their own instead of filing internal requests
• Helped set up Jenkins based CI/CD with Docker and Google Test and started to break legacy C++ services into smaller containerized pieces
• Built a React and FastAPI monitoring dashboard for station control systems so operators could see sensor data from more than twenty IoT nodes on one live screen.

Software Engineer Intern

• Built responsive React and TypeScript interfaces with modern state management and routing so the apps worked smoothly across desktop and mobile
• Improved reliability by hardening API clients with retries and timeouts and by adding web-vitals and custom events, so frontend issues were caught before reaching users