Most teams treat evals like a last-minute checkbox—ship first, panic later—but that’s exactly backwards. The real edge comes from treating evals as a continuous, evolving system from day one, not a static test suite. Because here’s the uncomfortable truth: LLMs don’t fail cleanly or consistently, and neither do your users. If you’re not constantly adapting how you evaluate, you’re basically flying blind—just with more features to hide it.

It's 2026, and We're Still Talking Evals

The 5xP Framework is a practical strategy that uses five targeted Markdown files (Product, Platform, Process, Profile, and Principle) to seamlessly align AI coding assistants with your project's architecture and business goals. By defining strict context boundaries, this framework drastically reduces prompt bloat and prevents AI hallucinations, moving developers away from unstructured "vibe coding" and closer to reliable, spec-driven development.

The 5xP Framework: Steering AI Coding Agents from Chaos to Success

Why Agents are Driving Software Development to the Cloud

Conversation with Mihail Eric on how agent-driven development is reshaping engineering work, faster iteration, new failure modes, and shifting team dynamics. Focus on validation, cost tradeoffs, and what breaks when code is mostly generated rather than written.

The Modern Software Engineer

This week’s Coding Agents Lunch &amp; Learn takes a more relaxed, interactive format. Instead of a single deep-dive talk, we’re hosting a Show &amp; Tell session led by Rahul Parundekar — a space for the community to share what they’re building, experimenting with, or thinking about in the world of coding agents. Expect a series of short, informal 5-minute presentations from participants, followed by open discussion. Topics can range from early prototypes and agent workflows to ideas, challenges, and lessons learned along the way. This session is less about polished demos and more about learning in public, exchanging ideas, and sparking conversations around where coding agents are heading. If you’ve been building (or just exploring), this is a great chance to share and get feedback.

Coding Agents Lunch and Learn: Show & Tell – Community Builds, Ideas, and Experiments

In this session, we’ll explore how coding agents can orchestrate a full MLOps pipeline, from data ingestion to deployed models, without manual intervention. Leo Walker will walk us through a real project built at the Stanford “LLM x Law” Hackathon, where agents were used to: <ol><li data-list="bullet">Pull and process data from external APIs</li><li data-list="bullet">Use LLMs for parsing and data cleaning</li><li data-list="bullet">Train classification and regression models</li><li data-list="bullet">Deploy model weights to Hugging Face</li><li data-list="bullet">Expose models via MCP servers for agent-to-agent interaction</li></ol> We’ll discuss how this approach shifts workflows from manual pipelines to coordinated agent systems, and what that means for the future of MLOps and applied AI. This session is ideal for anyone interested in: <ol><li data-list="bullet">AI agents in production workflows</li><li data-list="bullet">Automating ML pipelines</li><li data-list="bullet">Practical applications of LLMs beyond prompting</li></ol>

Coding Agents Lunch & Learn Session 9 : End-to-End MLOps with Autonomous Agents

After several sessions of Coding Agents Lunch &amp; Learn, we’re shifting gears into something more hands-on.In this interactive workshop, we’ll go beyond discussion and actually build. Together, we’ll walk through how to create practical AI “skills”, from defining the problem and structuring the approach, to testing, evaluating, and improving them in real time.A big theme we’ll touch on is First Principles Thinking, not just as a concept, but as something we can attempt to structure into usable, testable workflows. We’ll explore why some skills fail, how to scope them effectively, and how to introduce lightweight observability and evaluation so we can measure what’s actually working.By the end of the session, we’ll have:<ol><li data-list="bullet">Built a couple of simple, purpose-driven skills</li><li data-list="bullet">Defined evaluation criteria and tested them live</li><li data-list="bullet">Collected quick feedback signals (success rates, gaps, edge cases)</li><li data-list="bullet">Identified ways to iterate and improve</li></ol>This session is designed to be interactive, experimental, and a bit messy, in the best way. If you’ve been thinking about building reusable AI workflows or want to better understand how to evaluate them, this workshop is for you.

Coding Agents Lunch & Learn: Skill Building Workshop (From Idea to Evaluation)

🚀 Coding Agents Lunch &amp; Learn – Session 7This session will be a casual deep dive into the rapidly evolving world of AI coding agents — what they are, how they actually work behind the scenes, and why they’re becoming such a powerful shift in how software gets built.We’ll explore the emerging patterns behind modern agent systems, moving beyond simple “chat assistants” into more advanced, autonomous workflows. That includes how agents use tools, coordinate across multiple roles, plan tasks, and even maintain memory over time to improve performance.Some of the themes we’ll dig into:<ol><li data-list="bullet">The shift from single assistants to multi-agent systems working together</li><li data-list="bullet">How agents reason through tasks and break down complex problems</li><li data-list="bullet">The role of memory, iteration, and long-running processes in agent behavior</li><li data-list="bullet">How these systems interact with codebases, tools, and environments</li><li data-list="bullet">What this means for the future of development workflows</li></ol>This isn’t meant to be a rigid presentation — more of an open, collaborative session where we can unpack ideas, share perspectives, and connect the dots to how we might apply these concepts in our own work.Bring your lunch, join in, and come hang with the team 👋

Coding Agents Lunch & Learn, Session 7

<h2>There's a specific set of problems that show up after the demo:</h2>Agent decisions that look reasonable in isolation but compound into failures across turns. Multi-agent handoffs that silently lose context. Security models that scan prompts but miss intent. Retrieval that's fast in staging and unreliable in production. Pricing models that can't survive contact with finance.Ship Agents is a two-hour virtual conference where practitioners share the failures, fixes, and production patterns they've earned the hard way - from securing agent intent across multi-turn interactions, to building data architectures that keep retrieval reliable past the demo stage, to designing APIs that actually work for tool calling and MCP integrations.Speakers from major tech companies and early-stage startups. Every talk grounded in systems that are live and serving real users. If you're past the prototype phase and deep in the "why does this keep breaking" phase, this is your event.

Ship Agents: A Virtual Conference

The MLOps Community is where machine learning practitioners come together to define and implement MLOps.
Our global community is the default hub for MLOps practitioners to meet other MLOps industry professionals, share their real-world experience and challenges, learn skills and best practices, and collaborate on projects and employment opportunities. We are the world's largest community dedicated to addressing the unique technical and operational challenges of production machine learning systems.

MLOps Community

Home

Forum

Events

Content

People

Contributors

Albums

Clubs

Chat

Channels

Tools

Help

Login to the MLOps Community community to connect with others, attend community events, and more!

Login or Sign Up for even more

To unlock all parts of the community and get the best experience, complete your profile.