Sven Kratz Email & Phone Number

Kirkland, Wa, Us

We're building in social + generative AI, and hiring, learn more: https://easelapps.ai/

Santa Monica, California, Us

* Engineered IOS watch UI for biometric messaging app. Replaced rule-based activity recognition module with bayesian model trained from user data. 20% improvement in message recommendation accuracy.* Led development of 5 AR lenses for research into collocated gaming and AR communications.* Developed world-first CNN-based dog emotion recognizer for pet-based AR lens experience.* Trailblazed full stack (hardware and software) development of an artificial nose prototype, using ML to classify a multitude of household scents.* Authored and filed 37 invention disclosures. Apps and lenses deployed to hundreds of thousands of users.Technologies: Swift | Objective-C | React Native | JavaScript | TypeScript | Lens Studio | Python | PyTorch | NumPy | Pandas | AWS | GCS

Mountain View, California, Us

Spearheading UI/UX design and development of highly usable human-machine interfaces (HMI) for autonomous trucks while considering the needs of stakeholders such as engineers, truck operations, triage, customers and investors. * Designed and implemented a new autonomous vehicle UI for our trucks. Demonstrated improvement of the UX for internal-, investor- and customer-facing technology demonstrations. * Rearchitected legacy JQuery-based UI code to use React.js, Bootstrap.js and MobX for better modularity, maintainability and extensibility of Kodiak's UI software stack.* Optimized THREE.js/WebGL 3D vehicle and sensor visualization for improved performance and visual esthetics.Technologies: React | Three.js | MobX | WebPack | JavaScript | C++ | Bazel

San Diego , Ca, Us

HMI for Autonomous VehiclesDeveloped patent-pending audio-based warning HMI for autonomous vehicles, augmenting situational awareness for drivers and engineers of autonomous trucks in safety-critical situations.Trajectory PredictionEngineered machine learning models for trajectory prediction of other vehicles in critical traffic situations. This involved data extraction from vehicle dataset bags, dataset preparation, model training, automated benchmark development and integration in online vehicle stack.

Stamford, Connecticut, Us

Bootstrapped an offshore team of 10 AI engineers and researchers with a focus on commericalizing Harman X AI products. Showcased industry-first driver monitoring system at CES 2019.

Stamford, Connecticut, Us

Senior Principal Engineer in the HARMAN FX Team, the organization's most forward-looking group, executing advanced projects and reverse-engineering science fiction, every day. * Advanced Automotive Driver Monitoring System (DMS)Architected and implemented an advanced automotive driver monitoring system that is trained with best-in-class AI models that understand the driver's cognitive and emotional state. Our system allows combining AIs for vehicle control and driver understanding to create the best possible user interface for autonomous vehicles. Cornerstone of HARMAN exhibit at CES 2019.* Robotic Vehicle HMIDesign, implementation and evaluation robotic shape-shifting vehicle HMIs for level 3-5 autonomous vehicle control, to be deployed in upcoming high-end vehicle of major auto OEM.

* Activity Recognition using RF SensorDeveloped an activity recognition system that is able to accurately recognize activities of desktop office workers and retail clerks / customers using data from RF radar sensors embedded under the work surface.* Through-the-screen Robot Navigation UIOperating telepresence robots manually can be cumbersome. We developed a natural user interface for robot teleoperation that allows the driving of robots by simply clicking on a location in the camera feed. This eases the burden on the operator and makes the process of navigating a robot around a space much more enjoyable.* Thermal-Haptic Interactive SurfaceInvented the world's first interactive surface that uses resistive heating and liquid cooling to generate thermal feedback in form of thermal pixels. This provides user interfaces or interactive content with an additional thermal feedback channel. Due to its simplicity of the design and economies of scale afforded by liquid cooling, we are able to scale thermal-haptic interactive surfaces to very large surface areas and a high numbers of thermal pixels. * AirAuth: Gesture-Based Authentication SystemAirAuth is a gesture-based authentication system that uses in-air hand gestures tracked by a short range depth camera to perform user authentication. AirAuth supports personalized user password gestures and is resistant against smudge and shoulder surfing attacks.

München, Bavaria, De

Research assistant conducting research on user interface software and technologies for mobile devices. Teaching assistant for master-level courses in computer graphics, mobile interaction and interactive objects.Notable projects:- PalmSpace: gesture-based around-device interaction for mobile phones using depth camera and point-cloud tracking- ShoeSense: spatial UI for mobile and wearable devices using shoe-based depth camera tracking- Attjector: attention-following wearable micro-projector system

Redmond, Washington, Us

Architected and developed "Mayhem", an open source end-user programming environment for Windows and .Net. Evaluated the use of depth cameras for mobile interaction and 3D scan-and-print system.

Bonn, Germany, De

I worked as a researcher / PHD student at the Quality and Usability Group, TU Berlin, doing research into sensor-based mobile interaction under the guidance of Prof. Dr. Michael Rohs.Notable projects:- HoverFlow: mid-air interaction prototype for smart phones using IR sensing and machine learning to detect user gestures above mobile devices- Iphone Sandwich: Pressure-based two-sided interaction prototype- Semi-Automatic Zooming for Mobile Map Navigation: tilt-based Speed-Dependent Automatic Zooming (SDAZ) interface for navigating large maps on mobile devices. Developed dynamic model-based approach to SDAZ and invented semi-automatic UI modes

Redmond, Washington, Us

Evaluated touch input response on novel high-speed multi-touch pad using load-cell pressure sensor and robotic arm. Developed and integrated Kinect-based navigation system for mobile robot. Prototyped computer vision modules, X11 connectivity and Arduino modules for home automation end-user programming environment.

Aachen, North Rhine-Westphalia, De

Software developer for the ReXplorer mobile city game. Supported development on main game engine and developed camera-based gestural input used as a novel, main input mechanism for gameplay. http://hci.rwth-aachen.de/rexplorer

Phd., Computer Science

Self-Driving Car Engineer Nanodegree, Computer Science

Phd. (Candidate), Computer Science

Diplom (M. Sc. Equiv.), Computer Science

Abitur