Patrick Wieschollek

Machine Learning PhD | Senior Applied Scientist at Amazon

Autonomous Robotics / Computer Vision / AWS Cloud Infrastructure

As an Autonomous Navigation & Perception Specialist, previously working on Amazon Scout, with expertise in Machine Learning, Computer Vision, and Massively Parallel Computing, I implement core functionalities for next-generation robots at Amazon. My algorithms are optimized for embedded systems. As a certified AWS Cloud Solution Architect, I further design, develop, and implement complex cloud-based solutions to support internal Amazon projects.

Implementations

Numerical Solver in C++11

A header-only c++11 implementation of common optimization algorithms and TensorFlow, Matlab bindings using Eigen3.

The MIT License (MIT) C++
code

Image Viewer

A scientific image viewer written in C++ and Qt accelerate by OpenGL. It supports multiple viewports which are synchronized when dragging and zooming one image.

GPL-3.0 C++
code

Cluster-SMI

A cluster monitoring system created with Golang to get a quick overview over GPU usage. It is exactly like nvidia-smi but for multiple machines.

GPL-3.0 Go
code

InfoMark

A web service created with Go+Elm to manage university courses with exercise sheets. Students can upload their solutions for exercise sheets, which are automatically tested in Sandbox-Environments (using Docker&AMQP). This is used for courses with over 400 attendees at the university.

GPL-3 Go, Elm
code

Digital Music Stand

A web-based cross-platform music sheet viewer that support a programmable foot pedal to scroll to the next page during live performance. This app can run on a raspberry pi.

GPL-3.0 Go
code

TensorFlow Recipes

Implementation of several recent papers in TensorFlow.

Python, Cuda
code

TensorFlow Inference

Boiler-plate code to run saved models from TensorFlow directly in plain C, C++ and Go.

Apache-2.0 C, C++, Go
code