experiments.sh
██████╗ ██████╗ ██████╗ ██╗███████╗ ██████╗████████╗███████╗ ██╔══██╗██╔══██╗██╔═══██╗ ██║██╔════╝██╔════╝╚══██╔══╝██╔════╝ ██████╔╝██████╔╝██║ ██║ ██║█████╗ ██║ ██║ ███████╗ ██╔═══╝ ██╔══██╗██║ ██║██ ██║██╔══╝ ██║ ██║ ╚════██║ ██║ ██║ ██║╚██████╔╝╚█████╔╝███████╗╚██████╗ ██║ ███████║ ╚═╝ ╚═╝ ╚═╝ ╚═════╝ ╚════╝ ╚══════╝ ╚═════╝ ╚═╝ ╚══════╝
user@lab:~/experiments$ whoami
Shengguang Cui // AI Engineer & Researcher
user@lab:~/experiments$ ls -la | wc -l
11 experiments found
user@lab:~/experiments$
> Displaying 11 of 11 experiments found
experiment_01.log
🌐
Jan 2026 - Present[ACTIVE]
$ Personal Portfolio — This Website
// Personal Project
>Full-stack 12-module portfolio built with Next.js App Router, each page with a distinct themed design
>VTuber-style AI chatbot featuring a 3D VRM avatar with procedural animations, ElevenLabs TTS with audio-reactive lip sync, and Web Speech API voice input
>Integrates 10+ external APIs and services: FAL AI art generation, Tone.js virtual piano, d3-geo travel map, live sports data, etc.
>JWT-protected admin system, Supabase backend for photo gallery and AI art gallery, Vercel CI/CD
Next.js 16React 19Three.jsTailwind CSS v4SupabaseElevenLabs
experiment_02.log
📹
Jun 2025 - Sep 2025[IN_PRODUCTION]
$ Live Captioning & Translation System
// Industry Project - Videospace, Inc.
>Engineered a hybrid Java-Python pipeline on Wowza Streaming Engine to pipe resampled audio to STT services, reducing caption latency from 6s to 1s
>Implemented real-time multilingual translation and AI-powered highlights generation using OpenAI and Gemini APIs
>Deployed in-house Whisper, PaddleOCR, and LLaMA models on Lightning AI with GPU auto-scaling; evaluated LoRA and RAG strategies for continuous knowledge injection
PythonJavaWowzaSpeech-to-TextOpenAI APILightning AI
experiment_03.log
💬
Sep 2025 - Dec 2025[COMPLETED]
$ Message Integration & Processing Engine
// Industry Project - Videospace, Inc.
>Built a Slack integration module with Django models, views, and REST API endpoints to handle OAuth authorization and asynchronous message retrieval
>Architected an asynchronous Message Processing Engine using Celery to fetch messages, segment conversations into topic clusters, and generate structured LLM-powered summaries
>Enabled users to search and browse AI-generated summaries of relevant Slack discussions across workspaces
DjangoCeleryREST APISlack APILLMOAuth
experiment_04.log
👁️
Mar 2025 - May 2025[COMPLETED]
$ Hallucination Reduction in LVLMs
// Research Project - UCLA
>Proposed a contrastive decoding approach based on On-the-Fly Preference Alignment to steer LVLM outputs toward factual content
>Re-implemented Activation Steering Decoding (ASD) as baseline; benchmarked on LLaVA-1.5 with POPE and LLaVA-Bench
>Demonstrated the proposed method reduced hallucinations relative to ASD and other baseline decoding strategies
PyTorchLLaVAVision-Language ModelsContrastive Decoding
experiment_05.log
🔒
Jan 2025 - Mar 2025[COMPLETED]
$ Automated Privacy Testing for LLM through Fuzzing
// Course Project - Trustworthy AI, UCLA
>Built a privacy testing harness in Python to evaluate LLMs' PII extraction from HTML profiles
>Extended PROMPTFUZZ with privacy-oriented mutators (context injection, persona/role prompts, obfuscation, pseudocode) and HTML-aware templating
>Uncovered a +7pp lift in attack success rate (85% → 92%) on GPT-4o through iterative mutation and response analysis
PythonLLM SecurityFuzzingPrivacy Testing
experiment_06.log
🤖
Oct 2024 - Dec 2024[COMPLETED]
$ Real-time Swarm Mapping with Battery-free Robots
// Course Project - Computational Robotics, UCLA
>Proposed a system for real-time environment mapping using swarms of solar-powered MilliMobile robots under strict energy constraints
>Designed a multi-robot architecture integrating UWB localization, BLE low-power communication, and button sensors for collision detection
>Implemented Kalman Filter for state estimation and Artificial Potential Field algorithm for coordinated environment exploration
>Built a simulation testbed to validate mapping accuracy, power efficiency, and area coverage under real-world constraints
PythonRoboticsKalman FilterUWBArtificial Potential Field
experiment_07.log
🧠
Feb 2023 - Dec 2023[COMPLETED]
$ HeteroPruneFL — Federated Learning Framework
// Research Project - Duke Kunshan University
>Designed a heterogeneity-aware FL framework that assigns each edge device a client-specific subnetwork via importance-based pruning
>Introduced dynamic sparse training (prune–regrow) to adapt per-client network topology to local data within memory constraints
>Engineered a reproducible PyTorch stack with server–client orchestration and benchmarked on 4 datasets vs. 3 baselines, showing consistent accuracy gains
PyTorchFederated LearningNeural Network PruningDistributed Systems
experiment_08.log
🖼️
Mar 2023 - Apr 2023[COMPLETED]
$ Campus Image Generative Models
// Course Project - Advanced ML, CUHK-Shenzhen
>Collected and curated a CUHK-Shenzhen campus image dataset with preprocessing and augmentation
>Trained GAN, DCGAN, and diffusion model (with ImageNet pretraining) in PyTorch
>Conducted comparative study using direct and t-SNE/LLE visualizations, confirming diffusion outperformed GANs in realism and mode coverage
PyTorchGANDiffusion ModelsComputer Vision
experiment_09.log
📡
Feb 2023 - Apr 2023[COMPLETED]
$ Sensor Network Localization via Optimization
// Course Project - Optimization in Data Science & ML, CUHK-Shenzhen
>Formulated 2D sensor localization as 4 optimization problems: nonconvex feasibility, SDP relaxation, SOCP relaxation, and nonlinear least squares
>Implemented steepest descent, ADMM, and ADAM solvers with CVX benchmarking
>Extended to noisy scenarios and compared robustness across all methods
PythonConvex OptimizationSDPADMMCVX
experiment_10.log
🚗
Jul 2022 - Aug 2022[COMPLETED]
$ Federated Connected Automated Vehicles
// Research Assistant - Network Communication & Economics Lab, CUHK-Shenzhen
>Implemented Deep Q-Network for autonomous driving in the CARLA simulator
>Surveyed federated learning use cases in connected automated vehicle scenarios
>Explored integrating online learning with knowledge distillation into federated frameworks
PythonDeep Q-NetworkCARLA SimulatorFederated Learning
experiment_11.log
✈️
Apr 2022 - May 2022[COMPLETED]
$ Easyfly — Flight Database System
// Course Project - Database System, CUHK-Shenzhen
>Built a flight database system with SQLite and a Django web interface for search and booking
>Trained a decision tree model for flight delay prediction
PythonDjangoSQLiteDecision TreeFull-Stack
contact.sh
🤝
$ cat ./collaborate.txt
Open to Collaboration
Research opportunities, open-source contributions, or interesting AI/ML projects. Let's build something together.
$mail -s "Collaboration" lilil@ucla.eduPID: 5556UPTIME: 778dMEM: 17%LOAD: 0.19