Patrick Flickinger Email & Phone Number

Redmond, Washington, United States

Burbank, Ca, Us

JedAI: Generative AI, agents

Glendale, California, Us

AI Character Studio

Us

University of Washington:• Commencement Speaker, Dec 2022: University of Washington MS in Technology Innovation (GIX)• GIX Master's Launch Project sponsor: • Aircraft emission reduction • River plastic monitoring• Guest Lecturer: Embedded System Capstone. Conferences:• AMS (American Meteorological Society) Annual Meeting: abstract speaker • NOAA Workshop on Leveraging AI: speaker and AI workshop leader • MLDS (Machine Learning Data Sciences), Microsoft Research: speaker and workshop leader • Microsoft Research Location Summit: speaker • AI for Earth Summit: speaker and workshop leader (2016- 2020) • Bloomberg’s 2019 Data for Good Exchange: committee member and reviewer

Redmond, Washington, Us

• Architected and led the development of an aircraft engine monitoring, IoT edge ML execution pipeline and distributed devices, which correlate disparate streams of data and guides insights through actionable reports and real-time feeds. The system is in use by a major US airport and aims to eliminate much of the unnecessary 100,000+ metric tons of Co2 emissions per year.• Designed and developed a distributed, highly-scalable AI hosting platform for PAWS, a game- theoretic reasoning model that predicts future poacher behavior. The platform is integrated with SMART Spatial Monitoring tools and PAWS is now available to animal game parks worldwide.• Sourced and augmented data to assist in Microsoft’s cybersecurity response to destructive nation state activities. Insights are used by Microsoft President, Brad Smith to influence US policy.• Aided time-sensitive research into Covid-19 by leading the development of an AI research system that securely trains/executes AI models on highly secure, opaque medical datasets for Mass General Brigham researchers and other hospitals.• Mentored students at GIX (University of Washington/Tsinghua University) in the creation of crowd-sourced, citizen science solution to track plastic and debris in waterways. The result is a product that combines off-the-shelf hardware components with a novel object tracking and color model.• As member of the Microsoft-led disaster relief steering committee, helped design the Assessment Data Model (CDM) to break down communication barriers that plague relief organizations during disasters.

Redmond, Washington, Us

• Designed and developed the AI for Earth machine learning inference platform, which hosts the AI for Earth ML APIs that serve thousands of unique users per day. The pipeline handles long-running AI models and provides real-time feedback to its users.• Created container-based data science competition system in Azure Kubernetes that executes, and scores containerized ML submissions. Its novelty comes from its break from typical competitions in that rather than scoring submitted labels, the platform scores a submission’s performance on blind input data, thus greatly decreasing the chance of bias and tampering.• Architected and developed the processes, infrastructure, and user support software that gives scientists a way to “drop in” and containerize their models for execution in the AI for Earth Azure pipeline.• Built the initial AI for Earth “land cover mapping” AI service offering.• Architected an edge-based ML modeling pipeline camera trap images.

Redmond, Washington, Us

Aria is a highly scalable near real-time telemetry and analytics SaaS application on Azure consumed by multiple teams in Microsoft (Skype, Office, OneNote, Outlook, Xbox).• Led, redesigned, and rebuilt Aria’s health monitoring, which provides real-time analysis of Microsoft’s telemetry data. This was one of the first internal products created with a microservice architecture. The redesign has resulted in the reduction of alert latency from 24.5 hours to 12 seconds. The new pipeline is self-healing, automatically-scaled, and has a built-in recovery mechanism.• Implemented automatic pub/sub functionality with novel Azure “super-queues” and added purge features to the Aria funnel-based analysis product.

San Francisco, Ca, Us

AMPS is a modern pub/sub engine designed for scalable high-throughput, low-latency messaging that is required in real-time deployments, such as in financial services.

Redmond, Washington, Us

Microsoft StreamInsight, SSIS (SQL Server Integration Services), Microsoft Parallel Data Warehouse

Designed and created solutions for Traffic.com/Navteq, Coca-Cola, BioClinica, Inc., GSI Commerce, Inc., PalmOne, among others.