Basis Postdoctoral Fellowship, Perception & Model Discovery - Basis

About Basis

Basis Research Institute is a nonprofit applied AI research organization with two mutually reinforcing goals.

The first is to understand and build intelligence. This means to establish the mathematical principles of what it means to reason, to learn, to make decisions, to understand, and to explain; and to construct software that implements these principles.

The second is to advance society’s ability to solve intractable problems. This means expanding the scale, complexity, and breadth of problems that we can solve today, and even more importantly, accelerating our ability to solve problems in the future.

To achieve these goals, we’re building both a new technological foundation that draws inspiration from how humans reason, and a new kind of collaborative organization that puts human values first.

About the Vondrick research group

The Vondrick group studies computer vision and machine learning. By training machines to observe and interact with their surroundings, we aim to create robust and versatile models for perception. We often investigate visual models that capitalize on large amounts of unlabeled data and transfer across tasks and modalities. Other interests include scene dynamics, sound and language and beyond, interpretable models, and perception for robotics. Our group is part of the Visual Computing and Machine Learning ecosystem at Columbia.

About the Fellowship

This postdoctoral fellowship is a collaborative initiative between Basis Research Institute and Columbia University's Vondrick Lab. Supported in part by the NSF's AI Institute for Artificial and Natural Intelligence (ARNI), the fellowship centers on research at the junction of perception and model discovery. As a fellow, you will be a key contributor to our focused efforts to develop systems capable of autonomously discovering causal models by conducting experiments in the physical environment.

Research Focus

Our research aims to understand and automate the generation of scientific models. This initiative requires integrating several areas—probabilistic inference, active experimentation, and program synthesis—with initial efforts elaborated in our blog post. The ultimate objective is to create an AI system capable of understanding the operating rules of real phenomena through directed experiments, and to develop a comprehensive theory of the principles by which it operates.

The research environment is both structured and adaptable, with multiple avenues for scholarly contribution. As a postdoctoral fellow, your expertise can shape various aspects of the project. Examples include developing:

• Inference methods that integrate perception and causal model discovery into a joint process.
• Algorithms for learning neural representations of 3D scenes that capture both geometry and local/causal dynamics.
• Algorithms for active control and experimental design with implementation in robotic actuators.

Participation in this fellowship will deepen your foundations in multiple domains of intelligence, including perception, representation, and reasoning, while also allowing for a balance of focused research, academic exploration, and software development.

Who We're Looking For

We’re looking for PhD-holding researchers with the following traits:

• Exceptional Foundations: Deep technical foundations in domains like statistics, programming languages, machine learning, physics, or mathematics. Flexibility in learning concepts from unfamiliar disciplines is essential.
• Areas of Interest: We're particularly interested in expertise in one or more of the following:
  • 3D Computer Vision: Learned 3D representations, such as NERF models
  • Neural Simulation: Learned simulators and neural surrogate models
  • Deep Probabilistic Machine Learning: VAEs, diffusion models, normalizing flows, etc
  • Robotics: Model/simulation-based robotic control
• Research Track Record: Candidates should have a proven research history, demonstrated by publications in recognized venues like CVPR, ICCV, ICML, NeurIPS, UAI, PLDI or through significant technical reports or impactful software projects.

Core Responsibilities

• Research and Development: Lead research in computer vision and AI reasoning. Produce papers, code, and models that align with Basis and Vondrick Lab goals.
• Model Building: Develop and validate computational models for object recognition and causal reasoning through both simulations and real-world experiments.
• Code Tech Contribution: Integrate project with Basis' core reasoning technology
• Publication: Prepare findings for top-tier conferences and journals
• Mentorship: Mentor junior team members and graduate students

Role Details

• FT/PT: This is a full-time, time-bound fellowship lasting between 1 to 2 years.
• Location: Limited to the New York City area; this is an in-person position.
• Hours: While we prioritize in-person collaboration for its benefits to creative work, there is a degree of flexibility in your working hours. Expect to be available during certain set times each week for meetings, and be prepared to attend multi-day Basis-wide in-person events.
• Salary: $100,000 plus benefits

FAQs

How is this fellowship different from a Research Scientist position at Basis?

Differences

• Duration: The fellowship is time-limited, usually spanning 1 to 2 years, unlike the Research Scientist roles which are permanent positions.
• Academic Collaboration: Fellows will have co-mentorship from academic PIs.
• Emphasis on Publication: Fellows are expected to publish their research, aligning more closely with traditional academic roles.

Similarities

• Contributions to Core Technology: Like Research Scientists, fellows are expected to contribute code to Basis core technology. Basis will support Postdoctoral scientists without extensive software engineering experience to develop these skills.
• Research Funds: Fellows will have access to an annual allocation of research funds ($5,000) to be used for research expenses, skill development, conference attendance, and more.
• Integration with the Basis community: Fellows will participate in Basis events, on-sites, and other activities, and may collaborate on other Basis projects.

What happens after the fellowship?

• Postdoctoral Fellows may be offered a full-time Research Scientist position at Basis, contingent upon performance and organizational needs.

• Basis will support Fellows pursuing academic positions with mentorship and training regarding the academic job market.

Application Components

Please submit the following:

• A cover letter (no more than 2 pages) that includes:
  • A brief description of your PhD research and past accomplishments
• Your CV

Application close date

The close date has passed (1/3/24), and we are currently considering applications. New applications may still be submitted at this time, but we do not guarantee that they will be considered.