CV

Hi, here is a brief summary of my Resume, you can also click pdf to see.

Contact Information

Name Tingyu Zhang
Professional Title Master's Student in Computer Science | ML Research & Engineering
Email tingyu.z@northeastern.edu
Phone 206-741-8937
Location , Seattle, Washington

Professional Summary

Master’s student in Computer Science at Northeastern University Seattle (4.0 GPA) with a background in Applied Mathematics. Published researcher at ICLR 2026, with hands-on experience in ML systems, LLM agents, and production engineering.

Experience

  • 2025 - 2025

    Shenzhen, Guangdong

    Research Assistant
    Southern University of Science and Technology
    • Designed and trained a continual learning Multi-Armed Bandit (MAB) model with parallel optimization, formulating a preference-theoretic reward function analogous to RLHF feedback modeling; implemented training loop in PyTorch and TensorFlow with vectorized batch updates to improve convergence efficiency.
    • Modeled the influence maximization problem as an online combinatorial MAB over a stochastic graph.
  • 2022 - 2022

    Shenzhen, Guangdong

    Machine Learning Engineering Intern
    HAOCHENG TECH HOLDING
    • Engineered end-to-end ML training pipeline integrating Random Forest and Logistic Regression for material classification on large-scale tabular data, achieving 87.3% accuracy; applied TF-IDF vectorization and NLP preprocessing via NLTK and scikit-learn.
    • Designed PostgreSQL schema with B-tree and composite indexes on high-cardinality video metadata fields, achieving a 15× latency reduction (30 min → 2 min).

Education

  • 2025 - 2027

    Seattle, Washington

    Master of Science
    Northeastern University
    Computer Science
    • Machine Learning Systems
    • Deep Learning Networks
  • 2021 - 2025

    Shenzhen, Guangdong

    Bachelor of Science
    Southern University of Science and Technology
    Applied Mathematics
    • Linear Algebra
    • Statistics
    • Probability Theory

Skills

Programming Languages (Proficient): Python, Java, C/C++, JavaScript, TypeScript, PostgreSQL, MongoDB, R, MATLAB, LaTeX
ML & AI Frameworks (Proficient): PyTorch, TensorFlow, scikit-learn, LangChain, HuggingFace, Pandas, NumPy, Matplotlib
Developer Tools & Cloud (Proficient): Docker, Kubernetes, AWS ECS, RabbitMQ, Git, Spring Boot, React, Node.js

Languages

Chinese : Native speaker
English : Fluent

Interests

Research Interests: Reinforcement Learning, Multi-Armed Bandits, LLM Agents, Vision-Language Models, ML Systems

Certificates

  • ICLR 2026 Publication - International Conference on Learning Representations (2026)
  • arXiv Preprint - arXiv:2512.21925 (2025)

Projects

  • AI-Powered Job Search Agent with Ollama

    Developed autonomous job search agent using Ollama with Llama 3.2 model running locally, implementing RAG architecture with ChromaDB vector store for personalized job matching.

    • Engineered multi-agent system for resume tailoring, cover letter generation, and application tracking using LangChain and Python, reducing application preparation time from 2 hours to 15 minutes per job.
    • Built web scraping pipeline using Selenium and BeautifulSoup to aggregate job postings, integrated with SQLite database and Celery task queue, collecting 2K+ daily job listings.
  • Astronomical Image Classifier: ViT Fine-Tuning & LLM Agent

    Fine-tuned a Vision Transformer (ViT) for multi-class astronomical image classification using LoRA on a remote GPU server via SSH, achieving 90.4% accuracy while reducing trainable parameters by over 90%.

    • Built a supervised few-shot learning corpus of 500+ prompt-completion pairs with interpolation curve ground-truth labels via a prompt engineering pipeline.
    • Benchmarked against LeNet-5 trained on MNIST (60K images, 80.3% accuracy) as baseline.
  • Cloud-Native Microservices for Distributed System

    Architected cloud-native microservices on AWS ECS with Docker containerization, Application Load Balancer for traffic routing, and auto-scaling policies.

    • Integrated RabbitMQ message queues with Apache Commons Pool for connection pooling, implementing manual acknowledgements and retry logic to ensure reliable asynchronous processing.
  • SpringBoot FullStack Book Management Web Application

    Architected full-stack application using React 19 with TypeScript and Tailwind CSS for the frontend, Spring Boot with PostgreSQL for the backend.

    • Developed RESTful APIs with Spring MVC supporting full CRUD operations, including multipart file uploads for book covers with client-side image preview using the FileReader API.