Hi, I'm Veronica Zhao

Data Scientist

Data Analyst

AI/ML Engineer

CMU MS in AI Engineering graduate student and NYU Computer Science and Data Science undergraduate student with extensive experience in machine learning, deep learning, and AI research. Previously developed innovative ML solutions for financial documents at Tetrix, developed innovative ML solutions for disease progression prediction at NYU Langone Health, and developed innovative ML solutions for video virality at Google AI Studio. Currently actively looking for full-time internships from May 2026 to Aug 2026 and full-time jobs starting from Dec 2026 after I graduate from CMU. Happy to connect - if you are hiring and would like to schedule a brief call to let me do intros, don't hesitate to email me!

Get In Touch

01. About

I'm currently pursuing my Master's degree in Artificial Intelligence Engineering at Carnegie Mellon University with a focus on AI applications, computer vision, and NLP, and I'm interested in Data Scientists or Machine Learning Engineer internships for Summer 2026.

I am an aspiring student passionate about advancing my expertise in areas like distributed systems, machine learning, and data engineering through further academic research and internships.

I am proficient in Python (TensorFlow, PyTorch, Keras, Scikit-Learn, PySpark, LangChain, OpenCV), SQL, Java, JavaScript, R, data visualization tools (Tableau, SPSS), and MATLAB (less proficient comparing to the previous ones). I specialized in machine learning frameworks and data-driven decision-making processes. My industry internships have provided hands-on experience in implementing these technologies at scale.

I am interested in the intersection of distributed systems, machine learning, and real-time analytics, with a focus on developing solutions to tackle real-world challenges in enterprise environments.

Feel free to reach out - whether you're interested in collaborating on innovative research or exchanging ideas in tech or academia, I'd love to connect!

02. Experience

AI Engineer Intern

Tetrix

May 2025 - Aug 2025

Developed Python-based AI workflows for automated retrieval and processing of financial documents from investment portals
Established monitoring tools to detect failures and improve system robustness
Contributed to research on cutting-edge LLM applications for financial data analysis and portfolio management

Data Science Research Assistant

NYU Langone Health

June 2023 - May 2025

Processed 15 years of patient data using Python, R, and SPSS
Improved respiratory disease prediction models to 97% accuracy
Co-authored paper accepted by 2024 MLHC Conference

AI Studio Fellow

Google

August 2024 - December 2024

Led development of dual-input predictive ML model trained on 27,000+ videos
Designed adaptive two-stage cascade pipeline reducing MAE by 30%
Developed multimodal temporal-semantic feature extraction framework (R² = 0.97)
Engineered scale-invariant prediction system for 100K to 500M views

Software Development Intern

Huajun Technology

July 2024 - September 2024

Developed distributed file storage system with 95% uptime
Enhanced web server architecture increasing capacity by 400%
Built CRM platform managing 500+ clients and $10M+ transactions

Software Development Intern

Wanhua Chemical Group

May 2024 - June 2024

Optimized real-time production line anomaly detection with a custom CNN
Reduced detection latency by 40% and improved recall by 20%
Performed large-scale data cleaning for sensor-based predictive maintenance

Data Scientist Intern

CITIC Bank

January 2022 - June 2022

Consolidated 5 million+ financial transactions to build real-time analytics dashboards
Introduced multi-node ETL pipelines using Spark, reducing data processing time by 50%
Assisted in implementing fraud detection algorithms for credit risk management

03. Projects

Smart Refrigerator Management System

Developed a Docker-containerized MERN stack application for managing food inventory, tracking expiration dates, and minimizing waste through Al suggestions
Built and secured a RESTful Express + MongoDB Atlas API with JWT authentication, role-based access, and express-validator input checks; achieved 60 %+ test coverage with Mocha, Chai, c8
Designed a dynamic, responsive React/Tailwind UI with guest-mode logic, hot-reloading, and Framer Motion animations; integrated OpenAI API to generate personalized recipe suggestions based on fridge contents
Engineered a real-time notification system using Socket.io for instant updates on food expiration and recipe suggestions, and a logging system using Winston for error tracking and debugging
Implemented end-to-end CI/CD pipelines using GitHub Actions, automating tests and production deployment to the DigitalOcean platform
Wrote and maintained Docker + Docker Compose configurations for consistent multi-environment deployment (local and cloud), supporting front-end, back-end, and database containers
Utilized Agile methodologies for project management and Git for version control and code review
Presented the project at a local tech meetup, receiving positive feedback and interest from potential users

LinguaFusion: Multi-Source Linguistic Fusion for News Classification

Engineered a multi-dimensional linguistics-based feature representation system capturing syntactic, semantic, and contextual document characteristics
Implemented advanced NLP techniques including POS histograms, entity type fingerprinting, and embedding aggregation methods
Achieved a 3.32% accuracy improvement over industry-standard TF-IDF baselines on the 20 Newsgroups dataset
Maintained full interpretability and computational efficiency while approaching deep learning performance levels
Developed a modular pipeline architecture using scikit-learn, NLTK, and spaCy, enabling efficient feature experimentation

Kaggle Data Science Competition: Loan Approval Prediction

Processed raw data for a credit risk prediction model, handling missing values, outliers, and applying feature encoding and scaling techniques. Designed advanced feature engineering, creating income-to-expense ratios and interaction terms, improving model accuracy by 15%
Built predictive models using LightGBM and CatBoost, optimizing hyper-parameters through grid search and random search techniques. Applied ensemble methods, including stacking and blending, to enhance model robustness and predictive power
Evaluated model performance using cross-validation and ROC-AUC metrics, ensuring consistency and generalization across datasets. Iteratively improved the model through feature selection and tuning, achieving a competitive position in the top 15% (Silver Award) of the Kaggle leaderboard

Human vs. AI-Generated Text Detection: Veritas-AI

Designed and developed a robust AI text detection model with 99.77% accuracy, leveraging cloud-based frameworks to analyze linguistic patterns and distinguish between human and AI-generated content, thereby mitigating misinformation and enhancing content authenticity
Engineered custom pre-processing pipelines to tokenize, normalize, and embed text data, optimizing feature extraction for neural network compatibility and improving detection precision across diverse text inputs
Refined training algorithms by implementing dynamic learning rate adjustments and gradient clipping, reducing training times from 1361 seconds to 307 seconds per epoch and boosting model efficiency for real-time deployment
Integrated the detection model into a scalable cloud infrastructure using serverless architecture, ensuring seamless real-time content verification across platforms while maintaining high throughput and low latency

Automated Decision System Audit: Fairness in Comment Toxicity Classification

Conducted an in-depth audit of a comment toxicity classifier to identify and address demographic biases, utilizing advanced natural language processing models such as BERT, XLNet, and GPT to reduce unfairness by 25%
Designed and implemented custom pre-processing pipelines for text normalization and tokenization, improving the classifier's ability to handle diverse linguistic patterns and demographic variations
Developed and integrated sentiment analysis modules to enhance toxicity detection, enabling more accurate identification of harmful content and improving precision by 18% across various user groups
Fine-tuned model parameters and employed adversarial training techniques to ensure the classifier's robustness, achieving better generalization and fairer outcomes in toxicity classification

Applied Deep Learning: Chatbot Development

Designed and developed a multi-modal GenAI chatbot leveraging Vision Transformer, LSTM, and BioBERT models to automate lung health assessments, integrating advanced image and text analysis for accurate diagnoses
Engineered a custom image processing pipeline to analyze patient-uploaded lung images using Mean Linear Intercept techniques, enabling the chatbot to achieve 98% diagnostic accuracy in detecting pulmonary conditions
Implemented natural language understanding capabilities using BioBERT to interpret patient symptoms and medical histories, enhancing the chatbot's ability to provide personalized health recommendations
Optimized model integration and deployment within a cloud-based infrastructure, ensuring seamless interaction between image analysis and conversational AI components for real-time health assessments

Hi, I'm Veronica Zhao

Data Scientist

Data Analyst

AI/ML Engineer

01. About

02. Experience

AI Engineer Intern

Data Science Research Assistant

AI Studio Fellow

Software Development Intern

Software Development Intern

Data Scientist Intern

03. Projects

Smart Refrigerator Management System

LinguaFusion: Multi-Source Linguistic Fusion for News Classification

Kaggle Data Science Competition: Loan Approval Prediction

Human vs. AI-Generated Text Detection: Veritas-AI

Automated Decision System Audit: Fairness in Comment Toxicity Classification

Applied Deep Learning: Chatbot Development

04. Contact