Leveling Up: Tools and Techniques to Make AI Development More Accessible
Jan 11, 2024
auto_awesome
Sarmad Qadri, founder and CEO of LastMile, discusses the need for software engineering rigor in AI development and the democratization of AI. Topics include AI config, AI workbook, experimentation and collaboration with generative AI frameworks. The podcast also explores the retrieval-augmented generation system, AI config and semantic retrieval as AI development tools, the concept of orchestration in software engineering teams, ETL focus and vector databases in AI data, and the importance of proper tools for AI development.
Last Mile aims to provide software engineers with a tailored workflow that combines ML ops principles with familiar software engineering practices, bridging the gap between engineers and generative AI.
AI Config, an open-source tool developed by Last Mile, focuses on version control and standardization of generative AI models, simplifying the development and deployment of generative AI applications for software engineers.
Deep dives
Last Mile: AI Developer Platform for Engineering Teams
Last Mile is an AI developer platform designed for engineering teams. It focuses on prototype and productionized generative AI apps specifically for engineers, rather than just ML practitioners. The founder and CEO, Sarmad Caudrey, previously worked as a software engineering manager at Facebook, bringing practical experience to the company. Last Mile was inspired by the revolution of software engineers developing AI applications for the first time. The company aims to create a new developer workflow that combines ML ops principles with familiar software engineering practices. The goal is to provide software engineers with the tools they need to build generative AI applications and integrate them into their existing workflows.
The Need for a New Developer Workflow
The emergence of generative AI applications created an opportunity to rethink the developer workflow. Last Mile aims to bridge the gap between software engineers and generative AI by providing a tailored workflow that aligns with their existing practices. Unlike traditional ML engineers, software engineers may be less familiar with tools like Jupyter Notebooks and the ML ops pipeline. Last Mile focuses on simplifying the software engineering process and incorporates ML ops concepts that apply to generative AI. By providing a user-friendly interface and integrating with existing software engineering tools, Last Mile enables software engineers to comfortably develop and deploy generative AI applications.
AI Config: Version Control and Standardization
AI Config is an open-source tool developed by Last Mile that focuses on version control and the standardization of generative AI models. The tool allows users to control and manage prompts, model generation settings, and other important configurations in a version-controlled manner. It also helps address the challenge of managing multiple generative AI models from different providers by providing a unified API and syntax. AI Config enables software engineers to easily prototype and experiment with different models and configurations, while also facilitating the deployment and integration of generative AI applications into the production environment.
Challenges and Considerations for Generative AI Development
Developing generative AI applications poses unique challenges, including data extraction, model governance, and monitoring. Last Mile recognizes the need for reliable and responsible development practices. They emphasize the importance of proper documentation, transparency, and evaluation throughout the generative AI pipeline. Additionally, they highlight the importance of continuous improvement and the incorporation of feedback in the development process. Last Mile acknowledges the evolving nature of generative AI and the need for ongoing research and development to address emerging challenges in the field.
Sarmad Qadri, founder and CEO of LastMile, a startup building an AI developer platform for engineering teams. This conversation delves into key artificial intelligence and machine learning themes, focusing on injecting software engineering rigor into the development of LLM and GenAI applications.