
2024 BAIR Graduate Directory
Highlighting the latest Ph.D. graduates from the Berkeley AI Research Lab and their impactful contributions to AI.

Merge Vision Foundation Models via Multi-Task Distillation
Efficiently merge multiple vision foundation models into a single multi-task model through joint distillation.

Vision-Based Hand Gesture Customization from a Single Demonstration
Effortlessly design and customize hand gestures using vision-based technology from a single demonstration.

Moonwalk: Advancing Gait-Based User Recognition on Wearable Devices with Metric Learning
Innovative approach for user recognition on wearable devices using gait-based metric learning

Chain-of-table: Evolving tables in the reasoning chain for table understanding
Proposing a framework to enhance table understanding by training LLMs to outline their reasoning step by step and updating tables iteratively, yielding significant improvements on benchmark datasets.

Modular: Optimize and deploy AI models with MAX Engine and MAX Serving
Accelerate deployment of AI models using MAX Serving on Amazon EC2 with ResNet50 and EfficientNet models

Command-R: Retrieval Augmented Generation at Production Scale
Scalable generative model Command-R optimized for Retrieval Augmented Generation (RAG) and Tool Use at enterprise production scale.

Chess.com boosts performance, cuts response times by 71% with Cloud SQL Enterprise Plus
Chess.com achieves 71% reduction in response times by upgrading to Cloud SQL Enterprise Plus

Join our first Security Talks of 2024: Trends, gen AI security, and augmenting defenses
Exploring 2024 cybersecurity trends, gen AI security, and defense augmentation in Google Cloud's Security Talks event

A window into protein folding: Lowering the barriers for AlphaFold Inferencing
Streamlining protein folding with Vertex AI AlphaFold Inference Pipeline for biotech drug discovery

Driving Looker customer innovations in the generative AI era
Empowering Looker customers with AI-driven BI innovations for data analytics

How Livesport activates data and saves engineering resources with BigQuery and Dataddo
Maximizing data activation and saving engineering resources using BigQuery and Dataddo