YETIAN (Richard) LU Richard-luyt

Hi there, I'm Richard (Yetian) Lu 👋

🎓 BCS @ UWaterloo 2030.

🌱 Currently diving deep into Machine Learning algorithms and Backend Development

🚀 Architecting scalable backend microservices and RESTful APIs to drive data-intensive applications and AI-integrated systems.

🎯 Deepening expertise in backend system design and core data structures to drive high-performance SWE and MLE solutions.

A RAG-powered brain that handles information with precision and curiosity.

Dual-Searching Mode:
- 🎯 Strict: High-threshold vector retrieval for accurate citations.
- 🌿 Explore: Relaxed similarity for broader knowledge discovery.

Predicts the bbox of selected vehicles in the next frame. Trained on 3050 4GB

Implemented ResNet18 + LSTM on the KITTI dataset to achieve temporal consistency in object detection.
Developed an Async FastAPI backend to serve model inference as a scalable microservice.

Real-time code-contextual collaboration platform for seamless peer-to-peer mentoring.

Contextual Messaging: Enables users to initiate discussions tied to specific code lines via Monaco Editor integration.
SSE Real-time Pipeline: Implemented Server-Sent Events to deliver instant cross-user notifications with lower overhead than WebSockets.
Secure Auth: Leveraged Auth.js for robust session management and protected collaborative workspace access.

Robot Path Tracking Controller

Implemented an A* path planner and a Pure Pursuit tracking node in C++17, translating vehicle kinematic constraints and lookahead geometry into a robust trajectory-following algorithm.
Integrated ROS 2 TF2 library for spatial coordinate transformations within a Dockerized simulation environment, publishing real-time velocity and steering telemetry visualized directly via Foxglove.

Main website and controlling pannel

Currently working on the main website development. Using Fastify, Next.js, PostgreSQL and so on. Planning to integrate student and teacher managing system.