Machine learning dataset engineer for GPT scale ML project at OneSecondDelivery

Promising drone-delivery startup is hiring a Dataset and Machine Learning Manager to work remotely. OneSec is working on a groundbreaking GPT-scale machine learning project to make autonomous delivery drones. In this job, you will build the Datasets and Neural Networks that fly our autonomous delivery drones. These drones will then deliver millions of meals and other packages in the real world.

About OneSecondDelivery

We are creating a food delivery app that uses autonomous drones to deliver the food. We deliver everything 5 times faster and 100 times cheaper, thanks to small autonomous delivery-robots. The company was founded in 2017 by a leading expert on drone legislation and a successful e-commerce entrepreneur. The Swiss Federal Institute of Technology’s Vice Presidency of Innovation has recognized the large potential of this project and has supported us with the XGrant Prize and other resources.

Job Description

OneSec is building one of the largest machine learning projects in the drone industry. With our own photorealistic flight simulator, and a fleet of prototype drones, we have built one of the largest 3d perception dataset in the world, and we are building a large computer cluster of NVIDIA and Tenstorrent chips to scale up our machine learning to GPT-Scale.

Your job has 3 components:

• Neural network training and optimization
• Dataset management and generation
• Supporting our hardware team in building up our exaflop-scale machine learning cluster

Here are some of the Neural Networks you will be building:

-Delivery surface finder: A neural net that picks the optimal delivery surface within a balcony or backyard of a customer. Usually this will be a table or chair or a section of the floor that has enough clearance.

-OneSec Fruxel Depthmap: This is a convolutional neural network that takes camera images and generates a simple, but highly sophisticated vector representation of the space around the drone. Our pathplanner then uses this depthmap to plan a collision free path for the drone.

-Visual Odometry Model: Our control system always needs an accurate understanding of how our drone is moving around from frame to frame.

-Payload Positon tracker: While delivering the package, our control system needs to know how the package is sliding out of the payload bay. This model simply takes in the camera images and outputs the current location of the package.

-Delivery Clearance Model: This model checks the camera images to see if the scene is safe to approach. We don't fly into a balcony if there's a human or dog in it. And when we approach a restaurant to pickup a package, we make sure the scene is clear of humans, pets etc.

Requirements

• The most important thing is that you are excited about this project, and like the mission of our company.
• Highly knowledgeable about machine learning and neural networks. Both practical and theoretical.
• Experience with computer vision and convolutional networks
• Experience with PyTorch and TensorFlow
• Good computer skills, and general tech savvy: you will do many tasks like setting up recording devices to record datasets, creating small python programs for dataset labelling and management, post processing large image datasets etc.
• Computer programming experience in other areas is helpful