Mitigating AI Risk in Practice with Guardrails AI with Shreya Rajpal

In this episode of ODSC’s AiX Podcast, we’re joined by Shreya Rajpal, the co-founder and CEO of GuardrailsAI. Prior to this, she was the Founding Engineer of Predibase and a Senior Machine Learning Engineer of the Special Projects Group at Apple. Guardrails AI is an open-source platform developed to ensure increased safety, reliability, and robustness of large language models in real-world applications. In the conversation, Shreya reflects on her journey from AI research to startups and self-driving cars, emphasizing the growing importance of AI safety. She shares insights on real-world AI failures, such as Air Canada’s chatbot misinforming customers, and the broader implications of unreliable AI adoption. Key Takeaways: - AI Reliability is Crucial: AI adoption is often limited by risks like hallucinations, misinformation, and compliance issues. Guardrails AI helps mitigate these risks to make AI safer and more reliable. - Guardrails AI in Action: The open-source framework provides over 65 AI guardrails, allowing businesses to set validation checks on AI inputs and outputs, ensuring accuracy and safety. - Real-World AI Failures Highlight the Need for Guardrails: Examples like Air Canada’s chatbot misinforming users show how unreliable AI can lead to legal and reputational consequences. - Low-Latency & Scalable AI Risk Management: Guardrails AI operates with minimal overhead, allowing companies to deploy safe AI without sacrificing performance. - Upcoming Guardrails Index: A new benchmarking study will compare AI risk mitigation models, providing insights into best practices for AI safety. Topics Discussed: - Shreya’s background in machine learning, self-driving cars, and how she ended up co-founding Guardrails AI - A deep, technical overview of the open-source side of Guardrails AI - Common mistakes companies make when building AI without safeguards - Detecting bugs and failures in AI models, specifically generative AI - Case studies of companies that haven’t set up guardrails properly References and Resources Mentioned: 1. Connect with Shreya! - LinkedIn: https://www.linkedin.com/in/shreya-rajpal/ - Twitter/X: https://x.com/shreyar 2. Learn more about Guardrails AI: - GitHub: https://github.com/guardrails-ai/guardrails - Website: https://www.guardrailsai.com/ - Guardrails Hub: https://hub.guardrailsai.com/ - Guardrails Blog: https://www.guardrailsai.com/blog - Discord: https://discord.com/invite/kVZEnR4WQK This episode was sponsored by: Ai+ Training ⁠https://aiplus.training/⁠ Home to hundreds of hours of on-demand, self-paced AI training, ODSC interviews, free webinars, and certifications in in-demand skills like LLMs and Agentic AI And created in partnership with ODSC ⁠https://odsc.com/⁠ The Leading AI Builders Conference, featuring expert-led, hands-on workshops, training sessions, and talks on cutting-edge AI topics and tools, from data science and machine learning to generative AI to LLMOps Never miss an episode, subscribe now!