An introduction to Open Accelerated Discovery
summary
OpenAD is an open source framework that wires deep search, generative design, simulation, retrosynthesis and robotic prototyping into a single API/CLI/notebook stack so experiments are recorded and repeatable. The demo shows docs embedded with Llama 2 + Chroma DB, a LangChain front end, scalable execution with Ray and secrets via Dapper, molecule similarity search (Lucene/Elastic), GT4SD generation feeding retrosynthesis tools and options to send routes to RoboRXN or ChemSpeed. It routinely filters 40,000 candidates down to practical, synthesizable leads.
AI-written summary · about
chapters (8)
- 0:00Introduction to OpenADK
- 0:11What is Accelerated Discovery
- 3:03Four Key Technologies
- 11:53OpenADK Framework Design
- 14:59Demo Walkthrough
- 21:44Simulation and Quantum Computing
- 26:18Open Science and AI Integration
- 26:39Q&A on Quantum Computing
related talks
12mOpen source EDA for Machine Learning
16mOpenRAG: An open-source stack for RAG
40mAgentic Relevance Tuning With OpenSearch: Faster, Easier, and Higher Quality Search at Scale
14mAutomating Design Space Exploration Using Advanced Simulation Technologies
52mSerg Masís | AI Track | Compass AI & Data Summit 2025
57mStartup Showcase
01h 06mAutomated design space exploration and generation of AI accelerators
42mCreating the Engine for Scientific Discovery