Vitor Rodrigues

Gemini Model Deployment: infrastructure velocity and model performance. My team works across the model development stack to deliver improvements in deployment quality, performance, and release velocity.

Engineering Manager III (L7) in ML Infrastructure in Google's Core ML org, focusing on supporting the entire Experimentation Workflow: workload execution, resource management, model exploration and iterative experimentation. Delivering highly scalable ML platform for training (experimentation, orchestration, analysis) used by most of Google researchers/engineers/teams do develop ML models, from idea to model, model experimentation/iteration and model to production journeys. Building model exploration solutions that let researchers go from a trained model/checkpoint to interacting with it via an API endpoint and/or interactive UI in a matter of seconds. Abstracting resource management away from users, while building economical incentives to workload prioritization and execution. Improving interactive experimentation workflows end to end, with a major focus on usability and developer velocity, improving time to launch ml workloads by 10x. Developing and driving the vision and strategy for training solutions. Supporting our team members in their personal growth and career development. Pre-2020: Google Health, applying AI/ML on top of EHR data to deliver better health outcomes.

Technical advisor for the startups accepted into the Residency program at Campus London, cohort of 2019, focused on ML for Good, with strong focus on AI for Health.

As part of the founding team, I’ve focused on a variety of areas and responsibilities since company inception. * Establish technical infrastructure (GCP, Docker, github, Flask, React) and development practices (Code Reviews, Coding Best Practices, QA, Continuous Integration, Prod push) * Build initial prototype client app (React + Redux) that launched to beta users in first 90 days * Develop Flask based server app to power prototype and diagnosis algorithms * Design and implement Curai’s Knowledge Base, a repository of medical knowledge facts that power our algorithms: * Devised a concept representation that can represent both pre-coordinated and post-coordinated concepts from multiple sources (SNOMED, Dxplain, QMR, Wikidoc) * Provide APIs for computing similarities and inferring relationships between concepts and support for runtime evaluation of multiple medical knowledge graphs * Built pipeline and tools for data ingestion and curation from multiple sources * Lead onboarding of First Opinion (acquisition) into Curai’s tech stack. Issue triage, clinical staff support and migration of tech stack from on-premises to GCP.

As part of the Campus Expert Residency program, I participated in a 2-week immersive mentoring experience where I worked closely with startups, at Campus and/or their own office locations. I worked closely with the following startups: * Logispot (https://www.kookjelogis.com/): engineering, scalability, deployment, code development best practices, production monitoring. * AIM (http://getaim.co/): system design, API & integrations, hiring & recruiting strategy, interviewing hands-on training, security design. * Impressivo (http://impressivo.net/): product development lifecycle, UX testing and research, pretotyping. In addition, I held open office hours at Campus and gave a techtalk on "Lessons learned from building a chatbot", where I shared Engineering and UX lessons learned and best practices from building and launching @meet, Google's enterprise assistant bot focused on scheduling.

Tech Lead for @meet, an AI powered enterprise bot to help customers with their scheduling needs. Led team of 10+ engineers, UX, QA. Designed plugin-based architecture for enterprise bot development. Drove project planning and deliverables from early prototype to public launch. Previously: - TL for user context service in G Suite, powering new features in Inbox's Quick Reply - experimental projects in Machine Intelligence group, building apps to showcase machine learning. - Google+ home stream engineer, full-stack development, implemented dynamic multi-column layout. - Unified user model that powered ranking across multiple products. - Internal tools.

Software Engineer for Optim Development Studio.

Technical enabler for new IBM technologies, including DB2 pureXML, IBM Data Studio and IBM pureQuery. Responsibilities include plan and develop proofs of concept, demoing at conferences and customer locations, technical articles and support early customers product adoption.

Internship in DB2 Native XML Storage FVT (Functional Verification Test) team. Job responsabilities: Xquery functionality testing, infrastructure management, code coverage measurement. Skills: Xquery, SQL, DB2, AIX, VMWare, Intel Compiler

6 month internship in DB2 Everyplace Performance Team. Job responsabilities: performance and system testing. Development of testing infrastrucure. Skills: Java programming, XSLT, DB2 Everyplace Administration.