Dahong Luo
Software Engineer, ML Researcher & Builder
Computer Science @ UMD focused on audio processing, generative AI, large language models, and computer vision.
Computer Science @ UMD focused on audio processing, generative AI, large language models, and computer vision.
I'm a Computer Science student at the University of Maryland (Math minor, GPA 3.87/4.0) with hands-on experience across machine learning research, full-stack software engineering, and hackathon product development. I currently work on compact neural audio codecs for low-power devices, and I enjoy building practical systems that connect strong modeling with real-world deployment.
Honors Program, GPA 3.87/4.0
Expected Graduation: May 2026
Implemented a diffusion pipeline in PyTorch with a U-Net backbone to generate cat images and study denoising behavior across timesteps.
Built hybrid quantum-classical models using Qiskit feature maps and variational circuits to explore learning behavior on structured datasets.
Fine-tuned SAM2 on domain-specific segmentation tasks using PyTorch to improve mask quality on challenging inputs.
Trained and adapted a vision-language model stack using Unsloth + Llama to transcribe mathematical figures and preserve structural meaning.
Developed an embedding-conditional GAN architecture in PyTorch for controllable generation with improved latent conditioning.
Conducted research using LSTM for Sound Event Detection that doesn't rely on future data. Wrote paper as first author, implemented data processing pipeline, and conducted ablation studies.
Built a PyTorch-inspired deep learning framework using only NumPy with custom layers, AutoGrad-like backpropagation, and full Transformer architecture, achieving 96.2% accuracy on MNIST.
Designed an AI that autonomously learns to play Pacman using reinforcement learning. Engineered a fully functional Pacman game environment from scratch using Pygame with modular architecture for easy integration with AI agents. Developed a Deep Q Network (DQN) model with PyTorch.
Won "Best Digital Forensics Related Hack" at Hoya Hacks 2024. Developed a web app to pinpoint sound timestamps in videos for faster investigations. Built and trained a YAMNet-inspired audio classifier in PyTorch with a custom audio-to-image preprocessing pipeline.
No projects match your search.
Grand Prize · Built an agent-based ELMS copilot with multi-check RAG and grade-aware study recommendations.
1st Place (Elevance Challenge) · Developed multilingual medical data pipelines from OCR and retrieval to SQL reasoning.
1st Place (Digital Forensics) · Built an audio event detector and React/Flask investigative web app.
1st Place (Spark of Genius) · Created JarWiz for hybrid hand-gesture + voice-based computer control.
Co-presenter at UMD AIML Club on fine-tuning Llama 3.1 3B with Unsloth, SFT, and LoRA concepts.
Presented a 2-hour workshop for 40+ participants covering model training, optimization, and backpropagation.