YJJiang, Yuhang
01 — Projects

Projects

Jiang, Yuhang · 2021 to 2025
项目
02

Selected Projects

14 of 14
Filter
  1. 01

    Spec2Sim-Agent: Multi-Agent System for Automated Simulation

    Project page Demo video Code

    • Developed a multi-agent system using Google ADK and Gemini 2.5 Flash Lite to autonomously convert natural-language hardware specifications into verified SimPy simulation code.
    • Achieved an 87× reduction in development time (from hours to roughly 41 seconds on average) by automating the translation and verification workflow with high precision.
    • Designed a specialised three-agent architecture (Analyst, Architect, Verifier) that decomposes complex engineering tasks into structured analysis, code generation and rigorous validation.
    • Implemented a “self-healing loop” where the Verifier autonomously detects simulation errors and feeds them back to the Architect for iterative correction without human intervention.
    LLM & Agents11.2025
  2. 02

    kDEN-NSGA-II Scheduler for Flexible Job-Shop Scheduling

    Code

    • Developed a multi-objective FJSP solver based on enhanced NSGA-II, optimising makespan, load balance and idle time across machines.
    • Designed a hybrid initialisation strategy combining heuristic, chaotic and random generators for population diversity.
    • Integrated adaptive crossover mechanisms with kNN-based density selection, enabling dynamic operator scheduling and diversity-aware evolutionary search.
    • Built a modular benchmarking framework supporting multiple baselines, including standard NSGA-II, NSGA-III, SPEA2 and multi-objective ACO.
    Optimization06.2025
  3. 03

    LLM Health Assistant

    Code

    • Developed an AI-driven health consultation platform using LLM and RAG, supporting text and voice interactions with PubMed paper retrieval, plus a CLI-based database management system.
    • Designed a scalable four-layer architecture (Process Centric, Business Logic, Adapter Services, Data Services) for efficiency and security.
    • Used FastAPI for the backend, SQLite for storage, Pinecone for vector search, OAuth 2.0 with JWT for authentication, and PubMed API integration.
    • Built a responsive frontend with HTML, CSS and JavaScript for seamless text and voice input.
    LLM & Agents02.2025
  4. 04

    Heuristic NSGA-II for Flexible Job Shop Scheduling Problems

    Code

    • Developed an advanced NSGA-II algorithm with heuristic population initialisation for the Flexible Job Shop Scheduling Problem, focusing on multi-objective optimisation balancing makespan and workload.
    • Designed a modular project structure covering data preprocessing, algorithm implementation, batch processing, evaluation metrics and visualisation.
    • Conducted experimental validation on benchmark datasets, demonstrating improvements in scheduling efficiency.
    Optimization02.2025
  5. 05

    High Performance Computing for Grey Wolf Optimizer

    Code

    • Designed and implemented the HGT-GWO algorithm, incorporating global historical best positions and individual trend guidance, significantly improving convergence speed and outperforming traditional GWO on three benchmark functions.
    • Proposed a novel master-worker island parallelisation scheme, enabling independent subpopulation operations and reducing communication overhead through controlled synchronisation intervals.
    • Conducted experimental validation in Python, demonstrating superior performance over GWO on 15 test functions.
    • Developed a fully parallelised implementation in C, MPI and OpenMP, tailored for UNITN’s HPC cluster.
    HPC01.2025
  6. 06

    Augmented Reality-Driven Robotic Arm Control for Industrial Automation

    SUPSI Project Course

    Code

    • Conducted a comprehensive literature review on XR technologies in Industry 5.0, highlighting human-centric design, worker safety and data privacy [PDF].
    • Developed and tested a system combining YOLOv8 and FastSAM for accurate image segmentation and fingertip coordinate mapping.
    • Designed an AR interface using Microsoft HoloLens 2 for real-time gesture recognition, allowing intuitive robotic arm control.
    • Achieved efficient performance in industrial environments, mitigating hand occlusion using time-sharing processing and ensuring flexible task handling.
    • Demonstrated scalability through multi-mode operation, enabling both gesture-based and interface-based controls for part picking.
    Computer Vision09.2024
  7. 07

    Semantic Segmentation of High-Resolution Remote Sensing Images with Non-Local Attention

    Outstanding Undergraduate Graduation Project

    Code

    • Conducted comprehensive research on efficient and accurate image segmentation algorithms for complex remote sensing images.
    • Developed an encoder-decoder model with residual-weighted attention to enhance feature extraction and mitigate performance degradation.
    • Ran experiments on ISPRS Potsdam and Vaihingen, achieving F1-scores of 90.23% and 87.37%, outperforming models without attention mechanisms.
    • Demonstrated proficiency in convolutional neural networks, Transformer models and attention mechanisms for semantic segmentation.
    Remote Sensing05.2023
  8. 08

    Video Anomaly Detection Based on Unsupervised Learning

    • Reproduced HF²-VAD (ICCV 2021, SOTA), tested on the CUHK Avenue dataset; performance slightly exceeds the paper by 0.2%.
    Computer Vision12.2022
  9. 09

    Music Generation Toolkit

    Audio sample

    • Invited by the NLP lecturer to give a talk to the next cohort of students.
    • A collection of music generation models from recent years. Data formats include compound word and REMI; models are mainly Transformer variants including Transformer-XL and Vanilla Transformer, freely composable to generate music.
    Others12.2022
  10. 10

    Semi-Supervised Medical Image Segmentation

    • Reproduced DTC (AAAI 2021), tested on the 2018 Atrial Segmentation Challenge dataset.
    Computer Vision12.2022
  11. 11

    Visualization Platform for COVID-19 Focus Segmentation

    Code

    • Offline inference, real-time display of segmentation results, and NII file export.
    • Used UNet and BiSeNetV2 to segment COVID-19 lesions.
    • Used PyQt5 to visualise the NII image format.
    Computer Vision10.2022
  12. 12

    Rock-Paper-Scissors Finger Guessing on Jetson Nano

    Code

    • Everything developed on the Jetson Nano, including data collection, model training and testing.
    • Supports real-time data reading from the camera and automatically returns game results.
    Computer Vision06.2022
  13. 13

    Automatic Compensation of Agricultural Insurance via Blockchain and Remote Sensing

    College Students’ Innovation and Entrepreneurship Training Program

    • Judge the disaster situation of crops through remote sensing images (image segmentation).
    • Use blockchain technology to encrypt the insurance compensation process.
    Remote Sensing06.2022
  14. 14

    Music Generation Based on LSTM

    Audio sample

    • Used music21 for MIDI format conversion, and built a music generation model with LSTM.
    Others11.2021

Note — for project experience related to competitions, see my Kaggle profile.