Explore the innovative use of Large Language Models at AngelList in this 52-minute podcast episode featuring Thibaut Labarre. Gain insights into how AngelList leverages natural language processing expertise to refine news article classification for investor dashboards. Learn about AngelList Relay, a groundbreaking platform that automates parsing and offers vital insights to investors. Discover the challenges faced, including Azure OpenAI collaboration and rate limit solutions. Understand the strategic importance of prompt engineering and how domain experts are empowered for ongoing advancement. Delve into topics such as system trade-offs, OpenAI limitations, human-in-the-loop processes, and ethical concerns in AI implementation. Gain valuable knowledge about investing in off-the-shelf models, back-end technologies, and the future of AI automation in venture capital infrastructure.
Using Large Language Models at AngelList - MLOps Podcast #171
Overview
Syllabus
[] Thibaut's preferred beverage
[] Takeaways
[] Please like, share, and subscribe to our MLOps channels!
[] A huge fan of Isaac Asimov
[] Thibaut Labarre background
[] AngelList as an organization
[] AI sense of building
[] System trade-offs
[] OpenAI's limitation
[] Human in the loop
[] Classifying relevance
[] Fight for value
[] Added value
[] Exploring efficient ways to automate tasks.
[] Investing in off-the-shelf models
[] AngelList Relay
[] News article and investment document classification technology
[] Back-end tech
[] Prompt layer
[] Prompt layer as a living
[] Foreseeing no human intervention
[] Blocking hallucinations
[] Challenges
[] Investments in other models besides OpenAI
[] Integration with other models
[] Ethical concerns when
[] OpenAI breaking Prompts
[] Wrap up
Taught by
MLOps.community
