Tyler Cao

Aspiring software engineer with hands-on experience in machine learning models, backend workflows, and data processing pipelines. Strong Python proficiency and practical project experience building transformer-based models, data-driven systems, and backend services. Passionate about full-stack problem solving and collaborative technical development.

• Transformer-based Text Autocomplete, Python, PyTorch, Implemented an end-to-end language modeling system to learn next-token prediction from raw text. Built a custom tokenizer & vocabulary using regex to handle nuances in English text. Designed and trained a Transformer model with positional encoding and causal masking. Created a PyTorch Dataset/DataLoader pipeline for sliding window sequence prediction. Implemented training loop with AdamW optimizer, gradient clipping, and validation perplexity metrics. Added an interactive CLI for top-k word suggestions and temperature-controlled text generation., reproducible
• Data Cleaning & Analytics Script, Python, Developed a Python pipeline to clean and structure unformatted time-stamp logs for forecasting. Applied regex, Pandas, and logic to extract consistent entries. Prepared cleaned inputs for downstream model training or dashboard use.

• Team Lead - Data & ML project team, Diablo Valley College, Organized weekly coding sessions and walkthroughs on PyTorch, data pipelines, and model debugging., Guided new members through hands-on exercises and collaborative GitHub practices., Designed a tokenizer and vocabulary system in pure Python for custom English text processing., Transformer-based language model in PyTorch with positional encoding and causal masks for next-token prediction., Evaluated model effectiveness using perplexity metric and developed interactive tools for top-k next-word suggestions and text generation.
• Open Source Contributor, Vue & Nuxt UI Components Project, https://github.com/unovue/inspira-ui, Contributed to the development of a Vue/Nuxt-based open-source UI component library focused on building modern, animated, and reusable interface elements., Designed and refined modular, reusable components following Vue's component architecture, with attention to composability, consistency, and long-term maintainability., Improved documentation, API descriptions, and usage examples to enhance developer experience and reduce onboarding friction for new users and contributors.