Machine Learning Engineer on the Search Relevance team

• Improved quality of data used to train search models by implementing data quality checks for freshness and logical correctness

• Built offline Learning to Rank Feature Logging REST API and ETL

• Collaborated on Proof of Concept for Confluence Ask AI RAG feature

• Designed and developed pipeline to automate search relevance labeling using GPT-4 to replace human labeling

• Collaborated on model training pipeline to train XLM-RoBERTa as a cross encoder for reranking over semantic search layer

• Performed prompt tuning to improve accuracy of GPT-4 in human search relevance labeling task

Software Engineer on the Search Relevance team

• Built ETL and Tableau dashboard to measure and report customer search success rate and north-star search metrics

• Collaborated on offline job to replay historical search queries for use in optimal ranking signal weight selection

• Drove end to end development of new ranking signals in Confluence Search including feature materialization pipeline, feature serving APIs, shadow A/B test and rollout to customers

Undergraduate TA (Tutor) in the Cognitive Science Department for COGS 118A: Supervised Machine Learning Algorithms under Professor Zhuowen Tu.

I taught students the basics of Python, including the basics of numpy, matplotlib, and sci-kit learn, as well as the basics of supervised machine learning algorithms such as linear and logistic regression, SVMs, perceptrons, KNNs, and random forests. I also graded assignments and exams, and held office hours.

My project with Livongo involved creating data streaming services from MySQL to Elasticsearch using Apache Kafka as an intermediary.

I also integrated Kafka into Livongo's registration flow to provide real-time updates to data that is used by Livongo's microservices and stored in Livongo's Elasticsearch clusters.