We're Hiring! Foundation Models Intern

Full-time & Remote

Ever wondered what the meaning is in the songs of whales, in the chirping of the birds in your garden, or in the ultrasonic communications of bats? You’re not alone…

Earth Species Project (ESP) is a non-profit research lab and impact organization dedicated to using AI to solve these questions. We believe the exponential progress we’re seeing in AI offers new ways of looking at the world and expanding the ability of human beings to learn from other species. Our hope is that this will make a significant contribution to altering human perspective on how we relate to the rest of nature.

ESP partners with biologists and machine learning researchers at leading universities and institutions around the world, and we are honored to be supported by many forward-looking philanthropists and groups, including the Paul G Allen Family Foundation and the National Geographic Society.

As the ability of AI to generate fluent human communication has advanced, so too has the public and scientific interest in what the potential of this technology might be for the animal domain. ESP’s work has been widely featured in the mainstream and scientific press over the past three years, from The Guardian to the New York Times, TIME magazine, Science, Scientific American, WIRED and the Financial Times.

We have also published our work at leading machine learning conferences and journals including AVES (Hagiwara, ICASSP 2023), BEANS (Hagiwara et al., ICASSP 2023), BioLingual (Robinson et al., ICASSP 2024), BEBE (Hoffmann et al., 2023), and Science (Rutz et al., 2023).

PURPOSE OF THE ROLE

Join our team for an exciting internship focused on advancing the application of foundation models in the field of bioacoustics. You will have the opportunity to explore transfer learning from state-of-the-art audio foundation models to bioacoustic tasks. This position is fully remote. 

You will work closely with the foundation models team, with your primary point of contact being Masato Hagiwara.

RESPONSIBILITIES

Depending on your interests and experience, your work may include:

• Evaluating existing audio foundation models on bioacoustics benchmarks such as BEANS, BIRB, DCASE
• Training and fine-tuning foundation models on general-purpose audio data to improve downstream performance on bioacoustics benchmarks 
• Experimenting with quick adaptation methods such as SLM (Wang et al., 2023)to bootstrap a bioacoustics foundation model
• Collaborating with the research team on applying the foundation model to current ESP projects

EXPECTED OUTCOME

By the end of the internship, we aim to:

• Gain insights into the potential of audio foundation models in bioacoustics
• Release foundation model code and weights fine-tuned for bioacoustic applications
• Co-author a research paper based on the internship findings

REQUIREMENTS

• Currently pursuing a PhD degree in Computer Science, Electrical Engineering, or a related field
• Experience in training and evaluating neural networks
• Proficiency in programming with Python and familiarity with machine learning libraries.

PREFERRED QUALIFICATIONS

• Proficiency in deep learning frameworks (PyTorch preferred)
• Previous experience with bioacoustics data
• Experience in training and fine-tuning large-scale foundation models (BERT, T5, HuBERT, LLMs, etc.)
• Publication in leading signal processing or machine learning conferences (e.g., ICLR, IJCAI, ICML, NeurIPS, ICASSP, Interspeech)

COMPENSATION

Up to 10,000 USD a month, based on experience and qualifications

TERM

The internship is flexible, with a minimum commitment of 3 months full-time, with the possibility of extension on a part-time basis

HIRING PROCESS

1. Application Screening (cover letter and resume)
2. First Interview
   • Note: each interview will be either one or two of Masato Hagiwara, Marius Miron, and David Robinson
3. Written Exercise
4. Second Interview
5. Offer

At Earth Species Project is a fully remote team, we strive to create a culture where all can thrive and be our authentic selves is essential to our mission. We believe our success depends on our diversity and fostering an inclusive environment that supports creativity and innovation. As an organization, we are committed to promoting equity and belonging in our work.

We are committed to equal employment opportunities regardless of race, color, religion, gender, gender identity or expression, pregnancy, sexual orientation, marital status, ancestry, national origin, genetics, disability, age, veteran status, and criminal history, consistent with legal requirements. We encourage folks of all backgrounds and perspectives to apply.

If you require any accommodations, please email us at jobs@earthspecies.org, and we’ll work with you to meet your accessibility needs.

Please submit your application by clicking the link below. Applications are reviewed on a rolling basis, but we encourage you to apply as soon as possible. The position will remain posted on Earth Species Project’s website until it is filled.

APPLY HERE

‍