Unleashing the Power of Private GPTs with Skyflow’s Manny Silva
Partially Redacted: Data, AI, Security, and PrivacyShow Notes
Manny Silva, Skyflow’s Head of Documentation, joins the podcast to share his journey of tinkering with generative AI systems and building a private GPT trained on internal Skyflow documents. Manny discusses his first impression of ChatGPT, how he got interested in this space as a technical writer, and the non-obvious insights he gained along the way. He addresses common misconceptions about GPT, particularly regarding privacy and security. Manny explains the concept of creating a private GPT and explores the reasons why organizations would want to implement it. He provides valuable insights into effectively integrating a private GPT into existing workflows and systems, along with the challenges and considerations companies should be aware of. Manny shares best practices for training and fine-tuning a private GPT to ensure optimal performance and accuracy. He delves into the impact of his work at Skyflow and the enhanced productivity observed in the field. Finally, Manny looks ahead to future advancements and trends in the field of private GPTs and discusses their transformative potential in the realms of documentation, product launches, and marketing.Topics: When you first saw ChatGPT, what was your first impression? As a technical writer, how did you get so interested in this space and start tinkering with the Open AI platform and APIs? What are some of the non-obvious things you learned as you dove into this? What are some of the common misconceptions you’re seeing when it comes to GPT, in particular when talking about privacy and security? What’s it mean to create a private GPT and why would someone want to do that? How can organizations effectively implement and integrate a private GPT into their existing workflows and systems? What are some common challenges or considerations that companies should be aware of when building and utilizing a private GPT? What are some best practices and strategies for training and fine-tuning a private GPT to ensure optimal performance and accuracy? Can you describe what you built at Skyflow that leverages private GPT?\ What kind of impact are you seeing in terms of yours or other people’s productivity? Looking ahead, what advancements or trends can we expect to see in the field of private GPTs, and how will they continue to transform the way we work with documentation, product launches, and marketing? Resources: Privacy-First AI: Harnessing Snowflake and Skyflow to Customize GPT Generative AI Data Privacy with Skyflow GPT Privacy Vault
More from Partially Redacted: Data, AI, Security, and Privacy
In this episode, Sean sat down with Jack Godau to dive deep into the world of pseudoanonymization. They started by discussing Jack's career trajectory working with highly sensitive data and how that experience shapes his engineering mindset. Jack shared how pseudoanonymization differs from anonymization, explaining its value for maintaining data utility while complying with stringent regulations like GDPR. Jack also walked us through the challenges and key components of building a pseudoanonymization engine, including the complexities of handling re-identification risks, ensuring scalability, and optimizing performance for large datasets. He shared insights on the trade-offs between data protection and usability, and whether building these systems in-house is worth the investment for startups. Finally, they explored where the field is heading, especially as data privacy concerns continue to grow.
In this episode, Sean sits down with Ben Burkert, Co-founder and CTO of Anchor, to dive into the world of certificate management and internal TLS. We explore how certificates and TLS function, the inherent difficulties in managing internal TLS certificates, and why nearly every engineer has a horror story related to it. Ben also shares insights into how Anchor is addressing these challenges and making internal TLS certificate management simpler and more reliable. Key Topics:Understanding Certificates and TLS:Basics of how certificates and TLS work.The role of TLS in securing internal communications. The Challenges of Internal TLS Certificate Management:Why managing internal TLS certificates is so difficult.Common pitfalls and challenges engineers face. Engineer Horror Stories:Real-world examples of certificate management gone wrong.The impact of these failures on teams and organizations. How Anchor is Fixing the Problem:Anchor’s approach to simplifying internal TLS certificate management.Key features and benefits of Anchor’s solution. If you've ever struggled with internal TLS certificates or are looking for a way to avoid the pain altogether, Ben’s expertise provides a clear path to overcoming the challenges of certificate management with a modern, reliable approach. Resources:https://anchor.dev/https://lcl.host/
In this episode, we sit down with Ori Rafael, CEO and Co-founder of Upsolver, to explore the rise of the lakehouse architecture and its significance in modern data management. Ori breaks down the origins of the lakehouse and how it leverages S3 to provide scalable and cost-effective storage. We discuss the critical role of open table formats like Apache Iceberg in unifying data lakes and warehouses, and how ETL processes differ between these environments. Ori also shares his vision for the future, highlighting how Upsolver is positioned to empower organizations as they navigate the rapidly evolving data landscape.
In this episode, Sean Falconer is joined by Aubrey King, solutions architect and community evangelist at F5, to discuss the top 10 security issues for LLM applications. They explore critical threats such as prompt injections, insecure output handling, and training data poisoning, among others. Aubrey provides insights into why these issues arise, the attacks being observed, and the methods used to mitigate these risks. This episode is essential listening for anyone interested in the security of large language models and their applications.
In this episode, host Sean Falconer sits down with Eric Flaningam, a researcher at Felicis Ventures, to explore the fascinating world of data warehouses. They dive into the history, evolution, and future trends of data warehousing, shedding light on its importance. Key topics discussed include an overview of the article "A Primer on Data Warehouses," and the definition and key characteristics of data warehouses. They also cover the historical evolution and major milestones in data warehousing, the shift from batch processing to real-time data, and the convergence of data warehouses and SQL. Eric and Sean discuss the impact of unstructured and complex data, advancements in technology and their effect on data warehouses, and the technical architecture and components of a typical data warehouse. They share real-world benefits and use cases of data warehouses, common challenges in implementing and maintaining data warehouses, and future trends and the influence of AI and machine learning on data warehouses. For further reading, check out Eric Flaningam’s article, A Primer on Data Warehouses: https://www.generativevalue.com/p/a-primer-on-data-warehouses