🎓 M.S. in Computer Science @ University of Rochester
🧬 Working at the intersection of AI, single-cell genomics, and systems
🌎 Based in Rochester, NY
📧 yhu116@u.rochester.edu | yhu116@ur.rochester.edu
📞 (603) 866-4355
🔗 GitHub

I’m a computer scientist and researcher interested in how modern ML and foundation models can be used to understand complex systems — from disease-associated brain cell states to end-to-end AI music generation. I enjoy building things that actually run: reproducible pipelines, research code that other people can use, and small systems that behave like “real” tools rather than class demos.

Right now, my work spans single-cell foundation models (scGPT, Geneformer), distributed systems, and multimodal generative AI.

Research Assistant, University of Rochester Medical Center
Training Generative AI to Identify Disease-Associated Cell States (May 2025 – Present)

• Build end-to-end pipelines for brain snRNA-seq datasets from psychiatric and neurodegenerative cohorts (QC, normalization, HVG selection, batch correction, metadata harmonization).
• Apply scGPT-based reference mapping and cell-type annotation; generate latent embeddings with Scanpy / PyTorch on NAIRR GPU clusters (Slurm).
• Fine-tune single-cell foundation models (scGPT, Geneformer) for cell-type/state classification and disease trajectories; run ablations and baselines and visualize results with UMAPs and model cards.

• Languages: Python, Java, C, C++, OCaml, Rust, Swift, HTML5, CSS, JavaScript
• Data / Stats / Scripting: R, Stan, MATLAB, MySQL, PHP, Ruby, C#
• ML / AI: scikit-learn, PyTorch, TensorFlow, Hugging Face Transformers, GitHub Copilot
• Cloud & Tools: Git, SSH, Docker, Google Cloud Platform, Google Colab, AWS, Xcode, Visual Studio, VS Code, JetBrains IDEs, RStudio, JMeter, AMP
• Focus Areas: Machine learning, deep learning, NLP, single-cell modeling, data mining, distributed systems

An end-to-end multimodal music-generation system.

• Uses Whisper for speech-to-text, Llama-4-Scout for lyric creation, Gemini Flash 2.0 for lyric structuring, and YuE for full music synthesis (vocals + instrumentation).
• Provides a Gradio-based UI and a multi-model pipeline supporting quantized deployment across different GPUs.
• Implements API orchestration and custom model adapters, benchmarking generation latency (~13–9 minutes depending on GPU).

A teaching-oriented but fully functional DVCS.

• Implements Git-like operations such as commits, branching, merging, and remote syncing.
• Designed around information hiding and minimal interfaces for the core FileLog module.
• Written in Rust, with a focus on correctness and clear separation of responsibilities.

Using ML to study lifestyle-related physiological patterns.

• Preprocessed and analyzed 70,000+ physiological samples to study associations between body signals and smoking/drinking habits.
• Applied PCA for dimensionality reduction and trained SVM and XGBoost models to predict smoking/drinking status, achieving ~81% accuracy with supporting visualizations.

NLP for political tweet classification.

• Classified political tweets from seven Northern European countries.
• Built an NLP pipeline with feature extraction and LinearSVC, achieving 77% accuracy.

University of Rochester – M.S. in Computer Science
Rochester, NY • Aug 2024 – Present

• GPA: 3.95/4.00
• Graduate Tuition Scholarship (50% of full-time tuition cost)
• A’s in all core CS courses (NLP, Machine Vision, Parallel & Distributed Systems, Databases, Networks, etc.)

University of Rochester – B.S. in Computer Science
Rochester, NY • Aug 2021 – Dec 2023

• GPA: 3.96/4.00
• Merit-based Tuition Scholarship ($5,000)
• Dean’s List every semester; A’s in all CS and Math major courses

Research Participant, Department of Mathematics
University of Rochester (Jul 2023 – Aug 2023)

• Studied fractal-based metrics for forecasting retail time-series data.
• Computed a Discrete-S Energy metric to approximate Hausdorff dimension and compared it with forecastability scores and compression-based complexity measures.
• Evaluated Holt-Winters, Prophet, and LSTM models and analyzed correlations between discrete energy, complexity metrics, and RMSE/SD.

Teaching Assistant
University of Rochester (Sep 2022 – Dec 2023)

• Supported courses including Computer Organization and Mobile App Development.
• Held office hours, answered questions on course material, and graded assignments and exams.

Computer Science Intern & Web Group Leader
VISION X LLC, San Jose, CA (Aug 2020 – Oct 2020)

• Led a small team to integrate Cohere LLMs into client websites using Django and AWS, creating embedded AI chatbots and premium subscription features.
• Optimized backend infrastructure for scalability and responsiveness, improving customer interaction quality by ~30%.
• Received a recommendation from the CEO for leadership and problem-solving.

• Fine-tuning single-cell foundation models (scGPT, Geneformer) on brain snRNA-seq datasets.
• Experimenting with training tricks such as Smart-Freeze, mixture-of-experts, and CUDA 12.x migration on NAIRR/ACES GPUs.
• Exploring AI + creative tools, especially systems that combine language, music, and generative models.

If you’d like to collaborate, feel free to open an issue, start a discussion, or reach out by email.