JPL Postdoctoral: Edge Vision Language Models and Foundation Models

NASA Jet Propulsion Laboratory

2d•Onsite

About The Position

New ideas are all around us, but only a few will change the world. That’s our focus at JPL. We ask the biggest questions, then search the universe for answers—literally. We build upon ideas that have guided generations, then share our discoveries to inspire generations to come. Your mission—your opportunity—is to seek out the answers that bring us one step closer. If you’re driven to discover, create, and inspire something that lasts a lifetime and beyond, you’re ready for JPL. Located in Pasadena, California, JPL has a campus-like environment situated on 177 acres in the foothills of the San Gabriel Mountains and offers a work environment unlike any other: we inspire passion, foster innovation, build collaboration, and reward excellence. We are recruiting a Postdoctoral scholar to conduct research in edge Vision Language Models and Foundation models. The successful candidate will play a key role in advancing the deployment and capabilities of state-of-the-art AI on edge devices. See more information at https://ai.jpl.nasa.gov

Requirements

Ph.D. in Computer Science, Artificial Intelligence, Computer Vision, Robotics, or a closely related field.
Research expertise in one or more of the following: Vision Language Models, Foundation models, or embedded AI.
Experience deploying AI algorithms and models to embedded platforms (e.g., Nvidia NX, AGX).
Hands-on experience with fine-tuning techniques (LoRA, SFT) and working with multiple vision heads (segmentation, YOLO, open vocabulary object detection).
Experience with evaluation and benchmark pipelines, specifically involving LLM-driven evaluation.
Demonstrated record of peer-reviewed scientific publications.

Responsibilities

Conduct research in edge Vision Language Models (VLMs) and Foundation models.
Deploy complex foundation models to embedded platforms (e.g., Nvidia NX, AGX, and others).
Fine-tune foundation models using techniques such as LoRA (Low-Rank Adaptation) and SFT (Supervised Fine-Tuning).
Develop and integrate multiple vision heads for tasks including segmentation, object recognition (e.g., YOLO), and open vocabulary object detection.
Design, implement, and run evaluation and benchmark pipelines, including LLM-driven evaluation methodologies.
Publish research results in peer-reviewed journals and present results at scientific conferences.