The New Stack Podcast

Rob Skillington, co-founder and CTO of Chronosphere, discusses building internal tools and the challenges engineers face. He emphasizes the importance of understanding project abstractions and considering build or buy decisions. Skillington shares lessons from his experience at Uber, highlighting the importance of knowing the audience and customer base. He addresses the 'not invented here syndrome' prevalent in organizations like Microsoft and suggests younger companies explore external solutions. The conversation provides insights into Skillington's experiences and considerations in developing internal tools and platforms.

Jean Yang, founder of API observability company Akita Software, emphasizes that programming languages should be shaped by software development needs and data, rather than philosophical ideals. Yang, a former assistant professor at Carnegie Mellon University, believes that programming tools and processes should be influenced by actual use and data, prioritizing the developer experience over the language creator's beliefs. With a background in programming languages, Yang advocates for a shift away from the outdated notion that language developers are building solely for themselves.In this discussion on The New Stack Makers, Yang underscores the importance of understanding the reality of developers' needs, especially as developer tools have evolved into a full-time industry. She argues for a focus on UX design and product fundamentals in developing tools, moving beyond the traditional mindset where developer tools were considered side projects.Yang founded Akita to address the challenges of building reliable software systems in a world dominated by APIs and microservices. The company transitioned to API observability, recognizing the crucial role APIs play in enhancing the understandability of complex systems. Yang's commitment to improving software correctness and the belief in APIs as key to abstraction and ease of monitoring align with Postman's direction after acquiring Akita. Postman aims to serve developers worldwide, emphasizing the significance of APIs in complex systems.Check out more episodes from The Tech Founder Odyssey series:How Byteboard’s CEO Decided to Fix the Broken Tech InterviewA Lifelong ‘Maker’ Tackles a Developer Onboarding ProblemHow Teleport’s Leader Transitioned from Engineer to CEO Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

Docker CTO Justin Cormack reveals that Docker has been a go-to tool for data scientists in AI and machine learning for years, primarily in specialized areas like image processing and prediction models. However, the release of OpenAI's ChatGPT last year sparked a significant surge in Docker's popularity within the AI community.The focus shifted to large language models (LLMs), with a growing interest in the retrieval-augmented generation (RAG) stack. Docker's collaboration with Ollama enables developers to run Llama 2 and Code Llama locally, simplifying the process of starting and experimenting with AI applications. Additionally, partnerships with Neo4j and LangChain allow for enhanced support in storing and retrieving data for LLMs.Cormack emphasizes the simplicity of getting started locally, addressing challenges related to GPU shortages in the cloud. Docker's efforts also include building an AI solution using its data, aiming to assist users in Dockerizing applications through an interactive notebook in Visual Studio Code. This tool leverages LLMs to analyze applications, suggest improvements, and generate Docker files tailored to specific languages and applications.Docker's integration with AI technologies demonstrates a commitment to making AI and Docker more accessible and user-friendly.Learn more from The New Stack about AI and Docker:Artificial Intelligence News, Analysis, and ResourcesWill GenAI Take Jobs? No, Says Docker CEODebugging Containers in Kubernetes — It’s Complicated Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

In this episode, Stefano Maffulli, Executive Director of the Open Source Initiative, discusses the need for a new definition as AI differs significantly from open source software. The complexity arises from the unique nature of AI, particularly large language models and transformers, which challenge traditional copyright frameworks. Maffulli emphasizes the urgency of establishing a definition for open source AI and discusses an ongoing effort to release a set of principles by the year's end.The concept of "open" in the context of AI is undergoing a significant transformation, reminiscent of the early days of open source. The recent upheaval at OpenAI, resulting in the removal of CEO Sam Altman, reflects a profound shift in the technology community, prompting a reconsideration of the definition of "open" in the realm of AI.The conversation highlights the parallels between the current AI debate and the early days of software development, emphasizing the necessity for a cohesive approach to navigate the evolving landscape. Altman's ousting underscores a clash of belief systems within OpenAI, with a "safetyist" community advocating caution and transparency, while Altman leans towards experimentation. The historical significance of open source, with a focus on trust preservation over technical superiority, serves as a guide for defining "open" and "AI" in a rapidly changing environment.Learn more from The New Stack about AI and Open Source:Artificial Intelligence News, Analysis, and ResourcesOpen Source Development Threatened in EuropeThe AI Engineer Foundation: Open Source for the Future of AI Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

DockerCon showcased a commitment to enhancing the developer experience, with a particular focus on addressing the challenge of debugging containers in Kubernetes. The newly launched Docker Debug offers a language-independent toolbox for debugging both local and remote containerized applications.By abstracting Kubernetes concepts like pods and namespaces, Docker aims to simplify debugging processes and shift the focus from container layers to the application itself. Our guest, Docker Principal Engineer Ivan Pedrazas, emphasized the need to eliminate unnecessary complexities in debugging, especially in the context of Kubernetes, where developers grapple with unfamiliar concerns exposed by the API.Another Docker project, Tape, simplifies deployment by consolidating Kubernetes artifacts into a single package, streamlining the process for developers. The ultimate goal is to facilitate debugging of slim containers with minimal dependencies, optimizing security and user experience in Kubernetes development.While progress is being made, bridging the gap between developer practices and platform engineering expectations remains an ongoing challenge.Learn more from The New Stack about Kubernetes and Docker:Kubernetes Overview, News, and TrendsDocker Rolls out 3 Tools to Speed and Ease DevelopmentWill GenAI Take Jobs? No, Says Docker CEO Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

Florian Valeye, a data engineer at Back Market, discusses the evolving landscape of data engineering, focusing on Delta Lake. He explains how Delta Lake addresses challenges of data warehouses and data lakes, emphasizing asset transactions and scalability. Valeye's involvement with Delta Lake at Back Market optimized ETL processes and improved communication between data scientists and engineers. He also discusses his journey with Rust programming language and the future evolution of Delta Lake.

Liam Crilly, Senior Director of Product Management at NGINX, discusses the potential of WebAssembly and its promise of universal portability. WebAssembly is more mature on the client side, lacking sufficient runtimes and toolchains on the server side. Despite the need for a virtual machine, WebAssembly provides near-native compute performance through additional layers of optimization. The chapter explores the pipeline concept, server-side WebAssembly, and the power of Rust and WebAssembly for server-side web apps. WebAssembly functions as a standalone computer with potential for various tasks.

Jonathan Katz, principal product manager at Amazon Web Services, discusses the evolution of PostgreSQL, including its expanded uses and the competition it faced. He highlights the new features of PostgreSQL 16, such as logical replication, and the long-term development of Direct I/O. Katz also talks about the importance of community involvement and shares his experiences with PostgreSQL in Amazon RDS, emphasizing its role in simplifying application development.

The backlash against the practice of 'shift left' in developer security is explored, as organizations face challenges in addressing vulnerabilities. Business logic vulnerabilities and attacks targeting the frontend are becoming more prevalent. Collaboration between development and security teams is emphasized, along with using DORA metrics to assess the impact of security efforts. The podcast also discusses the benefits of threat modeling and the challenges faced by front-end developers in ensuring security.

The podcast discusses operational resiliency and the role of AI and automation in achieving it. It explores boosting developer productivity through automation but also highlights the challenges it may create for operations. The importance of automating repetitive tasks in incident response is emphasized, along with the benefits of AI-powered platforms in streamlining incident handling. The significance of confidence, psychological safety, and organizational recovery in taking risks is also discussed.