Pete Bunch Email & Phone Number

Mountain View

San Francisco

• Upgraded model training for inference of ad visibility.• Led the development of generic model training pipelines which could be used for a variety of different targeting types.• Investigated effects of GDPR on data and model quality.• Refactored controller and pricing systems, unifying separate frameworks for Brand and Direct Response advertising.

San Francisco

• Lead data scientist responsible for modeling, real time bidding and control systems fundamental to advertising campaigns targeting specific audience segments. Primarily technical contributions, but with a growing role in project leadership and wider engineering strategy.• Masterminded the design and implementation of a new framework to target and trade off between multiple customer performance goals, including derivation of optimal pricing equations, development of a new control system, and planning and coordination of software changes across the bidding stack, resulting in reliable and consistent performance and fewer unhappy customers.• Wrote new libraries for real-time model scoring, vastly reducing code complexity and adding a new layer of abstraction, facilitating faster product development, cleaner code deployment, and reducing both computational load and developer pain.• Upgraded the statistical methodology underlying the A/B test framework, introducing Bayesian inference and hypothesis testing, and providing a rigorous foundation for assessing hundreds of modeling experiments.• Won a company-wide hackathon for innovations in bid pricing strategies, and granted three trade secret awards.• Mentored two summer interns and a winning intern hackathon team.

San Francisco

• Evaluated new data sources and improved log parsing algorithms to extract double the volume of label data for the flagship search product.• Investigated numerous improvements to demographic modeling, including alternative multi-class classification models, methods for model calibration, and unsupervised approaches to discovering hidden classes.

Cambridge, Uk

Cheshunt, Uk

• Developed custom statistical models and learning algorithms for probabilistic forecasting• Implemented new procedures in SQL to run at scale on a Teradata data warehouse.

Cambridge, Uk

• Devised a number of new Markov chain Monte Carlo algorithms for machine learning with dynamical systems.• Taught undergraduates in mathematics, signal processing, and MATLAB.• Created a series of ‘Coffee Conference’ internal lab seminars.• Conducted undergraduates admissions interviews.

Fort Halstead, Uk

• Designed algorithms for in-service and prototype detectors.• Lead a £300,000 program of research into acoustic landmine detection technology.• Planned and executed experiments and equipment trials.• Attended and presented at meetings with electronics contractors and users.

Enfield, Uk

• Developed software in C and Python, adding support for a new protocol to client-deliverable code.

Doctor Of Philosophy (Phd), Statistical Modeling And Algorithm Design

Thesis: “Particle filtering and smoothing algorithms for challenging time series models.” • Formulated statistical models and algorithms.

Master Of Engineering (Meng), Information Engineering, 1St With Distinction

Activities and Societies: Sailing (Captain - Cambridge University Sailing Team - 2007-2008)• Two years general Engineering then two years.

International Baccalaureate, 45 Points