Teaching and Work Experience
Research Internship at Baidu Inc. (Aug ~ Sept 2023)
- Research on pavement abnormality detection in the Vehicle-Road Collaboration R&D Dept.
- Feature a high-recall PicoDet on mobile devices to filter “obviously negative” samples
- Incorporate PP-YOLOE on the cloud to predict the bounds and types of abnormalities
- Deliver a presentation on LLM prompting techniques and applications for intelligent traffic
- Receive the Outstanding Intern Award as a recognition of contribution
- Guide the students to grasp complex concepts through crafted examples
- Provide reasonable, directed, and customized support in coding assignments
- Unveil the reasoning behind designs to lay a solid foundation for mastering the ECE stack
Writing Assistant of “Principles of Research” (Feb ~ Jun 2022)
- Assist the students in appealing to the audience and crafting solid arguments in their writings
- Host group meetings during office hours to discuss common pitfalls and provide feedback
Research Projects
Supervisor: Sever Tipei Professor Emeritus, NCSA & UIUC iSchool
- Familiarize with codes in the full stack (LASSIE frontend, CMOD middleware, LASS backend)
- Use C++ to facilitate spatialization, reverberation, and modifiers by partials in a sound
- Implemente the LASS infrastructure to render multiple tracks (channels) for a partial
- Revise the LASSIE interface to pass extra information through the XML document
- Update the workflow in CMOD accordingly to achieve the desired sound effect
- Add “partial alignment” scrollboxes to the GUI with gtkmm and Glade
Supervisor: Yang Yang Associate Professor, Zhejiang University
- Implement Time2Graph and Time2Graph+ algorithms that model time series as shapelet transitions
- Incorporate Graph Attention Network with KMeans clustering to encourage shapelet diversity
- Accelerate shapelet extraction by nearly 100% with accuracy comparable to 10 benchmark SOTAs
Resources: Release pack Demo slides Paper Poster
Self-Supervised Semantic Segmentation on 3D Teeth Mesh (Jul 2021)
Supervisor: Zuozhu Liu Member, IEEE
- Preprocess 3D point clouds with trimesh and aggregate them into small patches
- Implement Bootstrap Your Own Latent and Point-Masked Autoencoder in Pytorch
- Fine-tune model parameters with a small amount of manually labeled data
Course Projects
Supervisor: Jian Huang Assistant Professor, UIUC
- Implement a 5-stage pipelined processor with forwarding and FSM-based hazard control
- Develop the generalized PLRU algorithm to fully parameterize the two-level cache
- Combine with branch predictor and prefetcher to achieve an IPC of 0.2697
Supervisor: Zuofu Cheng Teaching Assistant Professor, UIUC
- Design a GUI at pixel level with smooth “flowing notes” animation and real-time statistics
- Feature the SGTL5000 sound codec, a homemade I2S controller, and SDRAM-based audio playback
Resources: Executable image Demo video User manual
Supervisor: Steven Lumetta Associate Professor, UIUC
Zbigniew Kalbarczyk Research Professor, UIUC
- Develop various components (RTC, terminal, system call, etc.) of a kernel from scratch
- Feature extra functionalities (dynamic malloc, signal handling, etc.) for a fully functional OS
- Enter the final showcase and rank Top 5 among 50+ teams
Resources: Executable image Interactive demo User manual Tutorial
Supervisor: Klaus-Dieter Schewe Professor, ZJU-UIUC Institute
- Implement fibonacci heap and B+ tree to store patient information and preferences
- Stress-test the system to handle thousands of data points and generate weekly reports
Competitions
Supervisor: Thomas Honold Professor, ZJU-UIUC Institute
- Utilize grey model for short-term and multi-layer perceptron for long-term analysis
- Model investors with vision and risk tolerance parameters in the Capital Asset Pricing Model
- Achieve a 3200% investment return over 5 years and won the Finalist Award (top 2%)
Voluntary Work
Summer Social Practice in Jingdong, Yunnan (Jun 2022)
- As the deputy team leader, I conducted a two-day volunteer teaching program at Jingdong Minzu Primary School as per the tradition of mutual assistance between ZJU and the Jingdong people.
- In the following days, I researched the local food industry and hosted a two-hour online streaming session advertising the agricultural specialties.
- I also worked on media coverage and shared my thoughts through articles, videos, and songs.
Let it be code or music, do it like you never did before.