Looking for a job

Ryan Dang

@ryandang

Message

I build and evaluate machine learning systems, from LLM safety research to NLP models.

United States

Message

What I'm looking for

I'm looking to build and evaluate machine learning and LLM systems, especially work involving NLP, model reliability, AI safety, and measurable improvements to model performance.

At UC Irvine, I research LLM reliability and co-authored a peer-reviewed ESEM 2026 paper on identifying unreported threats to validity in research papers. I also built an LLM summarization web app with automated hallucination checking, reducing hallucinations by 10% across 100 test documents.

I've built NLP and transformer projects including a bilingual Vietnamese-English POS tagger, GPT-2 and BERT Yelp-review classifiers, and a PyTorch arithmetic transformer. My work combines model evaluation, preprocessing, optimization, and practical debugging to improve accuracy and training performance.

Experience

Work history, roles, and key accomplishments

Undergraduate Researcher

University of California, Irvine

Feb 2024 - Jun 2026 (2 years 4 months)

Co-authored a paper on LLMs accepted by ESEM 2026. Processed and analyzed research papers to assess LLM ability to identify unreported threats to validity, and experimented with prompting techniques.

LLM Prompting Research Data Analysis Paper Review

Alignment Research Incubator

University of California, Irvine

Oct 2025 - Dec 2025 (2 months)

Investigated AI safety and alignment concepts including interpretability and learning from human feedback. Built a web app with an automated hallucination checker, reducing hallucination rate by 10%.

AI Safety LLM Web Development Hallucination Detection Python

Machine Learning Engineer

Hugging Face

Apr 2025 - Jun 2025 (2 months)

Constructed a bilingual POS tagger with language-specific emissions and a vectorized log-space Viterbi decoder. Improved training accuracy from 0.87 to 0.91 by testing tagset granularity.

Machine Learning NLP Viterbi Python Vietnamese English

Machine Learning Developer

Hugging Face

Nov 2024 - Dec 2024 (1 month)

Fine-tuned GPT-2 with a custom classification head for Yelp review analysis, achieving 67% validation accuracy. Implemented BERT to improve accuracy to 75%.

GPT 2 BERT Fine Tuning NLP Python PyTorch

Education

Degrees, certifications, and relevant coursework

University of California, Irvine

Bachelor's Degree, Computer Science and Applied Mathematics

Grade: 3.8/4.0

Pursuing a Bachelor's degree in Computer Science and Applied Mathematics, and a Master of Science in Computer Science, expected to graduate in June 2028.