Kevin Haas Email and Phone Number

Building Google's foundation model development platform

Building the ML infrastructure for Waymo

I led the Deep Engines group at AWS, building the machine learning technologies used by AWS customers and internal Amazon teams. The focus areas for Deep Engines included distributed training, CV and NLP model performance optimizations, explainability, graph compilation, cloud training and inference on the cloud, ML on Kubernetes, and open source contributions to the Deep Java Library, Kubeflow, MXNet, PyTorch, and TensorFlow. The product portfolio included the Deep Learning AMIs, Deep Learning Compilers, Deep Learning Containers, SageMaker Clarify, SageMaker Debugger, and SageMaker Operators for Kubernetes.

Research and Machine Intelligence: I led the team responsible for the redesign of TFX, TensorFlow’s internal machine learning pipeline, onto a fully open-source software stack. I was also responsible for tf.serving, TensorFlow’s inference serving platform, and an engineering team for Sibyl, the precursor to TFX.Google Cloud Platform: I cofounded the Google Advanced Solutions Lab, enabling GCP ML customers to collaborate directly with the CloudML engineering teams. I was responsible for program strategy, customer vetting, building the engineering team, and solution delivery. I also led the CloudML pipelines team, which evolved into Kubeflow Pipelines. Google Assistant: I was responsible for the Assistant Core Infrastructure team and pivoted it from maintenance and production support to a focus on new foundational improvements for Assistant (microservices, CICD, on-device, tighter integration with Search).

Engineering leader for the Bing Silicon Valley campus, with a charter to make web search personally relevant by integrating content from social media into the search results. I managed the Bing Social infrastructure group, two Cortana experience teams, and an applied science team focused on improving the Bing relevance through social content. I also led the team that redesigned Bing’s user data management system to support additional internal customers, namely Cortana, Office365, and Windows.

Direct management of deliverables for multiple science, engineering, and program management teams. Led a joint science/engineering team to build a “web of concepts” entity knowledgebase generated from web content, Yahoo content, and 3rd party content. Impact included augmenting Yahoo Search with additional features centered on entity graphs of millions of entities, better query intent understanding, and enrichment of Yahoo content. I also managed engineering teams focused on Yahoo's Structured Web Initiative and Yahoo’s webmaster community, including Yahoo SearchMonkey, Yahoo Site Explorer, and Yahoo SearchBuilder.

Managed multiple development teams responsible for creating innovative solutions using IBM’s text mining technologies, including joint development projects with multidivisional research and engineering teams. Primary focus was the research and development of analytics projects centered on structured and unstructured information extraction, business intelligence, and virtualization.

Responsible for translating client business requirements into technical solutions, primarily for eCommerce and supply chain corporate business units. Responsible for a wide range of design activities from requirements analysis through solution design and implementation. Responsibilities also included implementation and project management tasks as well. Projects ranged in scope from $10MM to $50MM in total contract value; revenue generated from the larger projects exceeded $2B/quarter.