Website Amazon

Amazon 2026 Applied Science Internship – Frontier AI & Robotics (San Francisco)

Opportunity Overview

Attribute Details
Organisation Amazon (Frontier AI & Robotics Lab / Amazon Science)
Research Leadership Pieter Abbeel, Rocky Duan, and Peter Chen
Role Title Applied Science Intern – Frontier AI & Robotics
Job Location San Francisco, CA, USA
Work Commitment Full-Time (40 hours/week) for 12 continuous weeks (May – September 2026)
Compensation Range $157,300.00 – $212,800.00 USD annualised (Prorated for 12 weeks)
Education Level Enrolled in an Undergraduate degree program (Bachelor’s or above)
Benefits Relocation support, EAP, Mental Health Support, 401(k) matching

Role Scope & Research Architecture

The Applied Science Intern works directly with former Covariant founders and Amazon Science researchers at the San Francisco Frontier AI & Robotics (FAR) lab. The intern will advance the frontier of Vision-Language-Action (VLA) models, generative AI, and real-world physical manipulation

 

[Multi-Modal Tokens & World Models] ➔ [Sim2Real / Real2Sim Transfer] ➔ [End-to-End VLA Policies] ➔ [Physical Robotic Control & Manipulation]

Core Responsibilities

  • Algorithm Innovation: Develop novel, scalable algorithms combining Large Language Models (LLMs) and Generative AI for robotic perception and control.

  • Research Execution: Solve research problems on production-scale physical data, focusing on bimanual manipulation, scene understanding, and open-vocabulary grounding.

  • Cross-Functional Collaboration: Partner with researchers, roboticists, and software engineers to deploy foundational AI models onto physical hardware platforms.

  • Publication & Communication: Document experiments, perform rigorous statistical analysis, and author high-impact research outputs targeting top-tier AI/Robotics venues.

Candidate Eligibility & Technical Skills Matrix

Basic Qualifications

  • Academic Status: Currently enrolled in a Bachelor’s degree program or higher.

  • Age & Commitment: At least 18 years of age, available for a 40-hour/week, 12-week internship in San Francisco, CA.

  • Programming Core: Experience programming in Python, C++, or Java.

     

Technical & Research Skills Matrix

Domain Key Focus Areas & Technical Requirements
Robotics & AI Research

• Core Domains: Deep Learning, Reinforcement Learning (RL), Computer Vision, Motion Planning


• Advanced Frameworks: Multimodal LLMs, World Models, Image/Video Tokenisation, Vision-Language-Action (VLA) models
Robotic Control • Sim2Real and Real2Sim transfer techniques, Bimanual Manipulation, Open-Vocabulary Panoptic Scene Understanding
Software & Frameworks

• Primary Stack: Python, PyTorch, JAX


• Experience with distributed deep learning training and large-scale GPU infrastructure
Publications (Preferred) • Track record of publications in top-tier conferences (NeurIPS, CVPR, ICRA, RSS, CoRL, ICLR)

Selection Workflow & Evaluation Stages

[Resume & Publication Review] ➔ [Recruiter Phone Screen] ➔ [Technical Coding & ML Fundamentals] ➔ [Research Deep Dive / Presentation Loop]

  1. Resume & Publication Screening: Primary evaluation based on PyTorch/JAX coding skills, robotics/RL background, and publication record at top venues (ICRA, CoRL, NeurIPS, CVPR).

  2. Initial Recruiter Screen: Verification of 12-week availability, San Francisco relocation, and academic enrollment status.

  3. Technical Coding Interview: Assessment of Python data structures, algorithms, PyTorch ML primitives, and mathematical fundamentals.

  4. Research & System Architecture Loop: Presentation of previous research papers/projects to Amazon Applied Scientists, followed by deep dives into reinforcement learning, simulation tools, and model architecture design.

Upload your CV/resume or any other relevant file. Max. file size: 2 GB.