MLflow

Exciting news—MLflow 2.21.0 is live! ???This release includes significant features, enhancements, and bug fixes to improve documentation, GenAI prompt management, tracing & more. Here’s what’s new in MLflow 2.21.0: ?? Documentation Redesign:?MLflow documentation?is fully revamped with a new MDX-based website that provides better navigation and makes it easier to find the information you need! ?? Prompt Registry:?MLflow Prompt Registry?is a powerful tool that streamlines prompt engineering and management in your GenAI applications. It enables you to version, track, and reuse prompts across your organization. ? ??? OpenAI Agent SDK: MLflow Tracing now supports?OpenAI Agent SDK, a multi-agent framework developed by OpenAI. ?? FastAPI Scoring Server: The?MLflow inference server?has been migrated from Flask to FastAPI, enabling ASGI-based scalable inference for improved performance and throughput. ?? Enhanced Tracing Capabilities:?MLflow Tracing?now supports synchronous/asynchronous generators and auto-tracing for Async OpenAI, providing more flexible and comprehensive tracing options. ?? Explore all the new features & improvements:?https://lnkd.in/eZx-Q4Mj #opensource #oss #mlflow #linuxfoundation

Managing prompts in GenAI applications just got easier! With MLflow Prompt Registry, you can version, track, and reuse prompts seamlessly across your organization. ?? Check out this 1-minute video to see it in action! ?? ?? Learn more about MLflow Prompt Registry: https://lnkd.in/dTKFhURb #opensource #oss #linuxfoundation #genai #mlflow

?? Join Us for MLflow Office Hours on March 19! Come one, come all to our next MLflow Community Office Hours! This biweekly gathering is your chance to explore the latest in #MLflow, #GenAI, and #MLOps workflows while getting hands-on insights from experts. What’s in store: ?? Deep dives into GenAI and MLOps best practices ??? Implementation guidance and troubleshooting support ?? Exclusive previews of upcoming MLflow features ?? LLM tracking & evaluation strategies ?? Interactive Q&A – Share your screen, ask questions, and get real-time solutions from the MLflow team Whether you're just starting with LLM tracking or contributing to the latest PRs, this session is for you! Bring your questions and join the conversation. ?? Wednesday, March 19 ?? 8:00 AM PST / 11:00 AM EST ?? Register today: https://lu.ma/mlflow319 #opensource #oss #linuxfoundation

Stop Guessing, Start Tracing: Full GenAI Observability in One Line of Code ?? Struggling to keep track of your #GenAI calls across different frameworks and scripts? MLflow Tracing provides an easy way to record, organize, and compare LLM calls—including multi-step workflows—so you can debug and optimize your AI applications with ease. With just one line of code: ????????????.<????????????????>.??????????????(), you get full AI observability for OpenAI, Anthropic, LangChain, and more. ?? Get started in just 5?? minutes! Read the full guide:?https://lnkd.in/ePFp7pcH #opensource #oss #linuxfoundation #mlflow #tracing

Join us for our next MLflow Virtual Meetup on March 6 at 4PM PST (GMT -0800)! Stop fighting with model signatures and focus on your model logic! MLflow 2.20.0 lets you use Python's native type annotations to automatically validate inputs and infer model signatures. MLflow Software Engineer, Serena Ruan, will show you how this addresses some of the most common pain points in custom model development and simplifies your workflow with patterns for everything from simple data types to complex Pydantic models. Come see how these improvements make your MLflow experience more Pythonic and robust, with fewer runtime surprises. Bring your questions for our Q&A session! ?? RSVP -> lu.ma/8zidr8oe

?? MLflow Documentation Just Got an Upgrade! MLflow docs have a new home with a sleek, modern design focused on usability and readability. Here’s what’s new: ? Better navigation for finding what you need faster ?? Enhanced search functionality ?? Improved code examples & syntax highlighting ?? Responsive design that works well on all devices Explore the new and improved documentation and let us know your thoughts! ?? ?? Check it out: https://lnkd.in/ePkQbQws #opensource #oss #linuxfoundation #mlflow #machinelearning

A nice way to leverage open source: We use MLflow's OTel instrumentation for DSpy to send traces to Langfuse (YC W23). We've recently leaned into the OpenTelemetry standards for LLM observability after being initially hesitant -- and the feedback we've received has been great. This example of inter-operability is a nice case study for how open standards benefits the entire ecosystem. cc Steffen Schmitz who's implemented our OTel backend 3 weeks ago!

MLflow 2.20: Debug GenAI apps without leaving Jupyter! ?? MLflow Tracing lets you capture inputs and outputs of intermediate function executions—perfect for debugging GenAI applications. With over a dozen integrations, MLflow makes it easy to generate traces without modifying your existing code. ?? Now, you can view the MLflow Trace UI directly in Jupyter notebooks—no more tabbing out and breaking your workflow. Stay focused, debug faster! Try it out & see the difference in your workflow. Read more here ?? https://lnkd.in/gyN8tZMd cc Daniel Lok #opensource #oss #linuxfoundation #mlflow #jupyter #genai

MLflow's open source tracing capabilities now power LangFuse observability for DSPy applications, creating a seamless path for developers to monitor their #LLM applications. ? LangFuse uses MLflow's OpenTelemetry-compatible tracing system to provide observability for DSPy — the framework for algorithmic optimization of LM prompts. This integration showcases how MLflow's flexible architecture enables ecosystem interoperability without requiring explicit partnerships. What makes this possible: ?? MLflow tracing exports standard OpenTelemetry-compatible traces ?? Our sophisticated tracing implementation for DSPy captures the complete execution flow ?? Simple configuration redirects traces to any OpenTelemetry backend LangFuse uses MLflow's tracing infrastructure to capture and visualize DSPy execution flows. By leveraging MLflow's OpenTelemetry compatibility, traces can be routed to the LangFuse platform. The same tracing data can route to MLflow UI or any OpenTelemetry-compatible system. This interoperability demonstrates the value of MLflow's commitment to open standards and flexible architecture. Our work building robust tracing for complex systems like DSPy — which captures the nuanced execution flow of LLM optimization — is now benefiting the broader ML ecosystem. ?? ?? Learn more: https://lnkd.in/enYqCrFy #MLflow #DSPy #LangFuse #OpenTelemetry #MLOps #Observability

?? Passionate about AI/ML? Ready to shape the future of MLOps? We’re now accepting applications for the MLflow Ambassador Program—a unique opportunity to lead, connect, and grow in the AI/ML space! As an MLflow Ambassador, you’ll join a select group of technical leaders who: ? Share expertise through blogs, tutorials, and speaking engagements ? Connect directly with MLflow’s core team & fellow ambassadors ? Build personal authority in the MLOps and AI space ? Lead meetups & foster thriving AI/ML communities ? Gain early insights into MLflow’s future roadmap Spots are limited—apply today! ?? https://lnkd.in/gWZ8wTa9 Learn more about the program ?? https://lnkd.in/eKHZYppa #mlflow #linuxfoundation #mlops #opensource #oss