SENIOR DEEP LEARNING ENGINEER
Work with our team to design and implement cutting-edge deep learning and computer vision systems and algorithms for Geopipe products. You will have the opportunity to apply cutting-edge research in machine learning, computer vision, and computer graphics to geospatial data. You will implement, in Python, core deep-learning/computer vision and (inverse-)procedural modeling algorithms, with varying levels of oversight.
LOCATION
New York, NY
Burlington, Vermont
EMPLOYMENT TYPE
Permanent
What You’ll Do
● You will work closely with our cofounders, Thomas and Christopher, to understand
high-level product vision and customer requirements.
● You will have the opportunity to apply cutting-edge research in machine learning, computer
vision, and computer graphics.
● You will implement, in Python, core deep-learning/computer vision and (inverse-)procedural
modeling algorithms, with varying levels of oversight. These algorithms rely on mathematical
techniques from linear algebra, probability, and geometry.
● You will use Git to manage your source code.
● You will deploy your code on our cloud platform, to test it on large scale geospatial datasets
(and ultimately generate 3D content that will be available to customers). This will require you
to interact with remote machines via a Unix shell.
● Your contributions will help ensure the success of an early-stage startup.
● Your metrics for success will be based on meeting key R&D milestones.
● Your path to success involves an enthusiasm for the problem and product, a willingness to
learn new skills as needed, and an ability to modularize complex implementation tasks into
manageable components and execute on each of those components in turn.
What We're Looking For
Necessary:
● A problem-solving mindset and an enthusiasm for learning new skills are more important
than any particular prior knowledge or experience.
● Effective communication and collaboration with the cofounders is a must.
● You must convincingly demonstrate an understanding of how to design, train, and debug
artificial neural networks.
Flexible:
● We use a variety of neural network architectures and deep-learning algorithms in-house. We
expect to see expertise in one or more of the following: fully-convolutional neural networks,
transformer networks, GANs, weak supervision, reinforcement- and imitation-learning,
novel architectures (graph-convolution, PointNet, BSPNet, etc.)
● A mathematical background covering some subset of linear algebra, probability, multivariate
calculus, geometry, and numerical methods will be invaluable.
● Most of our existing ML code is in TensorFlow and Keras.
● Prior experience with modern C++ will allow you to better interface with other portions of our
data ingestion and product pipeline.
● We use Git for managing our source code. Prior experience with Git is best, but a background
with Mercurial or other DVCSs is an acceptable substitute.
● Our development and production machines all run some flavor of Linux or macOS. Comfort
with some flavor of Unix shell environment (e.g. bash) will be helpful.
● We are agnostic to your choice of development environment; however, we need a convincing
demonstration that you will be able to self-support in the environment of your preference.
Our Values
As a team, we are proud to share these core values:
● Collectively, developing and supporting intellectual integrity and diversity (“You gotta
have all your eigenvectors”)
● Pursuing cutting edge research that allows Geopipe to achieve competitive positioning
● Supporting individual learning and growth, especially facilitating collaboration and open
and active communication, while empowering contributors with autonomy and trust.