Peter Frick Email & Phone Number

California, United States

Mountain View, California, US

- Fine-tuned and and optimized multiple LLMs for unstructured extraction to achieve up to 20% increases in field accuracy, a 5x reduction in inference time, and 2 filed patent applications- Led a team of 5 scientists and 2 engineers in the design and implementation of a document intelligence service orchestrating 5 ML models and supporting 5.5 million.

Mountain View, California, US

Trained and deployed seven transformers-based information extraction models serving for 4 new countries and reducing the average response from 15 minutes to 4 seconds for supported customers

Santa Clara, CA, US

- Developed data pipeline to reveal dynamic trend insights from billions of text articles, using custom query tools, statistical analysis, and visualization- Collaborated with engineering to translate high-value, ad hoc data science scripts into a tested and CI-integrated library and served to internal non-technical users via dashboard

Dallas, TX, US

Dallas, TX, US

• Formulated data strategy for marketing campaign using clustering and statistical inference, resulting in 19% improvement in key metric
• Built pipeline using Spark and python to scan topics from one year of the entire English Twitter firehose, filter out spam, monitor for significant events, and generate interpretable visualizations
• Optimized an internal queue processing protocol using NLP, regression, and simulation, resulting in a 4% increase in customer engagement
• Implemented and tuned a neural-network based sentiment classifier on a GPU server, achieving 80% accuracy across domains. Applied sentiment model to social data to successfully identify products with low/high public.

San Francisco, CA, US

• Conceived NLP pipeline for social analysis as a consultant for startup, msg.ai
• Trained LDA model to learn tweet topics from dataset of ~100,000 industry-specific tweets
• Identified potential clients that significantly over/under-performed their competition in social customer service through bootstrap resampling

Stanford, CA, US

• Designed data pipeline to statistically analyze experiments measuring millions of DNA reads
• Defined and championed key scientific objectives through technical presentations, cross-disciplinary collaborations, and grant applications
• Directed experimental team to achieve a 10-fold increase in signal-to-noise

Nashville, Tennessee, US

• Defined novel experimental metrics for predicting cancer drug responses resulting in multiple publications, including two in Nature Methods
• Composed R scripts to automatically extract data from biomedical images (>100 GB), perform statistical model fits, and visualize results
• Mentored two undergraduate National Cancer Institute research fellows in experimental design, data analysis, and statistical modeling; both entered STEM graduate school program
• Conducted mathematical modeling to predict dynamics of cancer drug resistance using distribution parameters of time-series experimental data

Doctor Of Philosophy (Ph.D.), Systems Biology

Bachelor Of Science (Bs), Physics