Astrophysicist, AI (Postdoctoral Research Fellow) (IS-1330-11)

Application Deadline: 22 November 2024

Department: Smithsonian Astrophysical Observatory

Employment Type: Full Time

Location: Cambridge, MA

Compensation: $81,963 - $106,549 / year

Description

OPENING DATE: October 23, 2024
CLOSING DATE: November 22, 2024
TYPE OF POSITION: Trust Indefinite (Non-Federal)
DIVISION: Office of Director
LOCATION: Cambridge, MA
AREA OF CONSIDERATION: This position is open to all eligible candidates.
What are Trust Fund Positions?Trust Fund positions are unique to the Smithsonian. They are paid for from a variety of sources, including the Smithsonian endowment, revenue from our business activities, donations, grants and contracts. Trust employees are not part of the civil service, nor does trust fund employment lead to Federal status. The salary ranges for trust positions are generally the same as for federal positions and in many cases trust and federal employees work side by side. Trust employees have their own benefit program, which may include Health, Dental & Vision Insurance, Life Insurance, Transit/Commuter Benefits, Accidental Death and Dismemberment Insurance, Annual and Sick Leave, Family Friendly Leave, 403b Retirement Plan, Discounts for Smithsonian Memberships, Museum Stores and Restaurants, Credit Union, Smithsonian Early Enrichment Center (Childcare), Flexible Spending Account (Health & Dependent Care). 
Conditions of Employment

• Pass Pre-employment Background Check and Subsequent Background Investigation, as required.
• Complete a Probationary Period if applicable. 
• Maintain a Bank Account for Direct Deposit/Electronic Transfer.
• The position is open to all candidates eligible to work in the United States.  Proof of eligibility to work in U.S. is not required to apply.
• Applicants must meet all qualification and eligibility requirements within 30 days of the closing date of this announcement. 

OVERVIEW

INTRODUCTION
The Smithsonian Astrophysical Observatory (SAO) is at the forefront, internationally, of the scientific exploration of the universe. SAO combines its resources with those of the Harvard College Observatory to form the Harvard-Smithsonian Center for Astrophysics (CfA). The CfA is the best-known astrophysics center in the world. Its programs range from ground-based astronomy and astrophysics research to space-based research, the engineering and development of major scientific instrumentation for space launch and use in large ground-based facilities, and research designed to improve science education. The research objectives of SAO are carried out primarily with the support of Government and Smithsonian Institution funds, with additional philanthropic support.  Government funds are in the form of Federal appropriations or the form of contracts and grants from other agencies. In contrast, Institution funds are available to SAO through grants from the Institution's Restricted Funds, Special Purpose Funds, Bureau Activities, Business Activities, and non-Federal contracts and grants. 

SUMMARY
The purpose of this position is to participate in the development of methods to leverage Large Language Models (LLMs) to enhance scientific research and to combine them with representation learning methods for astronomical data (images, spectra, light curves, X-ray event files), to extract meaningful correlations between astronomical literature and astronomical data, through the use of contrastive learning approaches. The candidate will join a team of multidisciplinary researchers, including astronomers, natural language processing experts, and computer scientists working at the intersection between astronomy and artificial intelligence. 

MAJOR DUTIES

1. Conduct scientific research in fields of generative artificial intelligence, including large language models, representation learning, variational inference, and contrastive learning, collaborating with a multi-disciplinary team of researchers, including astronomers, computer scientists, and software engineers.
2. Contribute to the compilation of a multi-modal dataset for the purpose of self-supervised learning, including data from the Chandra X-ray observatory, NASA’s Astrophysics Data System, and other optical and infrared surveys
3. Formulate new approaches for the training of deep neural networks (transformers, variational auto-encoders, etc.) to create low-dimensional representations, and use the resulting representations for a set of downstream tasks including regression, classification, and contrastive learning.
4. Interact with Natural language Processing experts to in