Robert Giaquinto Email & Phone Number

San Francisco, Ca, Us

Building foundation models for the 3D world.

Seattle, Wa, Us

Applied research in machine learning and NLP, largely focused on training large language models for text, code, and structured information. - Worked on Amazon Q Developer (formerly CodeWhisperer) and Bedrock large language modeling (LLM) services.- Lead scientist in pretraining models for database interaction, including the design of a multi-task learning strategy combining structured and unstructured data.- Trained and evaluated tokenizers, lead data engineering efforts, and expanded evaluation of LLMs to numerous tasks and over 100 benchmark datasets.- Additional proof-of-concept projects incorporating cutting-edge techniques --- like multi-task learning, retrieval, assessing model robustness, in-context learning, contrastive learning, continual learning, and distributed training for scaling transformer-based models.

Minneapolis And St. Paul, Minnesota, Us

Research interests: deep generative models with applications in information retrieval, natural language processing, and image modeling.Advisor: Arindam Banerjee.

San Jose, Ca, Us

- Developed sequence-to-sequence models for Adobe's Sensei AI email marketing products.- Presented findings to internal audience of researchers on time-series and rare-event prediction methods.

Malibu, California, Us

- Machine learning research on an Intelligence Advanced Research Projects Activity (IARPA) research program for integrating human and machine forecasts.- Developed a new model to augment human forecasting of geopolitical, macroeconomic, and health events.- Our model supports collaboration between human forecasters by simplifying debates into two-sided arguments and identifying the top arguments on each side.

Toronto, On, Ca

- Scaled topic model algorithms to large corpora of legal texts.- Topics extracted from texts facilitated fast searching for relevant legal documents matching search queries posed by users.

Minneapolis And St. Paul, Minnesota, Us

- Built an automated system that extracts and shares key sections of doctor's notes with hospital patients.- Transformed unstructured rich text files from doctor's notes using natural language processing into a structured dataset.- Key sections of text were extracted using a semi-supervised classification algorithm, which incorporates hundreds of thousands of unannotated doctor's notes in the learning process.

Northfield, Mn, Us

Doctor Of Philosophy (Ph.D.), Computer Science

Master’S Of Science In Computer Science, (Machine Learning, Natural Language Processing)

B.A., Mathematics, Statistics