Autodata: Automatic Data Scientist
2026
AI agents that function as data scientists, iteratively building high-quality training and evaluation datasets. Agentic Self-Instruct converts inference compute into better data. Paper · Blog
News
A timeline of launches, research, and milestones.
Muse Spark Agents
2026
Drove research to develop self-improving agents that can recursively improve model's agentic capabilities — SoTA on MCP-Atlas, frontier on Toolathlon. Natively multimodal reasoning with tool-use and multi-agent orchestration. Muse Spark · Muse Spark 1.1 · Eval Report
Meta AI reached 1 billion MAU
2025
Improved Meta AI's multilinguality, enabled roll-out to 12 languages and 40+ countries. Blog · News
Launched voice mode and photo editing in Meta AI
Connect 2024
Launched updated Llama 3 model for voice mode. Improved Planner for photo editing with multimodal inputs. Blog
Launched Llama 3 on Meta AI
2024
Developed Meta AI's online RL with Mixture of Judges, improving reasoning, instructions following, safety. Blog
Launched Meta AI with Llama 2
2023
Debut Meta AI into Family of Apps. Created Orchestrator for tool use. My talk at Connect
Meta AI Few-Shot Learner (FSL)
2021
Developed FSL for detecting new forms of harmful content across 100+ languages. Blog
Training AI to detect hate speech
2020
Built RL-based framework to E2E optimize hate speech classifiers. Blog