Postdoctoral Scholar

4 days agoSalary: 90K - 90KResearch Jobs by Stanford Institute for Human-Centered AI

Stanford Institute for Human-Centered AI

Stanford HAI is Stanford University’s interdisciplinary institute for AI and data-science research, education, policy, and practice.

Company intelligence

Stanford, California, United States

About Stanford Institute for Human-Centered AI

Launched in 2019, Stanford HAI advances human-centered AI through interdisciplinary research, education, policy work, open scholarship, grants, and public-facing AI measurement initiatives.

View jobs by Stanford Institute for Human-Centered AI

Skills

Aws Computer Vision Data Governance Deep Learning Human Feedback Human-In-The-Loop Machine Learning Human Subjects Research Interactive Machine Learning Machine Learning Model Calibration Model Deployment Multimodal Learning Neural Network Noisy Supervision Privacy-Preserving Machine Learning Speech Recognition User Study Weak Supervision

About the Role

You will develop human-in-the-loop methods that use coach feedback as supervision for model adaptation. You will integrate transcript, audio, and video data while meeting consent, privacy, and regulatory requirements. You will build performance-monitoring infrastructure and lead a user study evaluating coaching efficiency and fidelity with and without AI support.

Requirements

Hold a PhD
Demonstrate extensive experience with modern deep neural network techniques
Demonstrate the ability to design, train, and deploy large-scale deep-learning models
Possess expertise in computer vision, speech recognition, or multimodal learning
Have experience with real-world technology deployment
Have experience with human-in-the-loop machine learning or a technical foundation in human feedback, weak or noisy supervision, or interactive machine learning systems

Responsibilities

Design methods that convert coach interactions into supervision for model adaptation
Determine how transcript, audio, and video data can be used consistently with consent, privacy, and regulatory requirements
Develop methods for multimodal signal integration and learning from sparse, noisy feedback
Build infrastructure to monitor model performance, calibration, and variation across contexts and populations
Design and execute a user study of instructional coaching with and without AI support
Evaluate coaching efficiency and fidelity and interview coaches about tool use

Benefits

USD 2,000 research and conference allowance
Relocation support up to USD 3,000 for international relocation, USD 2,000 for cross-country relocation, or USD 1,000 for west coast relocation outside the Bay Area

Hiring Process