Last week, we were joined by the co-founders of Guardrails AI to unpack guardrails for LLM applications: what they are, why they are important, and how to thread them through the AI lifecycle. Guided by the AI Guardrails Index, we explored techniques to handle jailbreak protection, PII stripping, and hallucination suppression.
Arcee AI just dropped AFM-4.5B—another Small Language Model (SLM) and the opening act of a new Arcee Foundation Model family. Packed with 4.5 billion parameters, AFM-4.5B is tuned for real-world enterprise workloads and slim enough to run everywhere from edge phones to massive GPU clusters. Join us as CTO Lucas Atkins unpacks the training pipeline—spanning 6.58 T tokens, distributed Spectrum fine-tuning, mergekit alchemy, and rigorous compliance—and shares what’s next for the AFM lineup. We have lots of questions about the details!
​Typedef recently released Fenic, an open-source, opinionated, PySpark-inspired DataFrame framework for building AI and agentic applications. If it delivers, analysts will juggle embeddings, summaries, and toxicity scores as effortlessly as today’s numeric columns. Tune in to discuss how to extract structured data from unstructured data, reading Markdown files into DataFrames, and explore where Fenic fits alongside best-practice tools for AI Engineers today!
RAG eval: what I learned about semantic similarity vs relevance
What is the role of smarter chunking in RAG applications? 🚀 In this AI Makerspace bootcamp project, Chris pits a “table‑of‑contents” (TOC) chunker—and Cohere‑powered re‑ranking—against a vanilla RAG pipeline, then grades everything with RAGAs’ LLM‑as‑judge metrics. Chris explores how different chunk sizes affect performance, and studies when semantic similarity outshines plain relevance (and vice‑versa).
Cohort 6 Highlight: Polarion AI Assistant by Shraddha Patil
Polarion AI Assistant implements classic RAG into chatbot that turns natural-language questions into precise, step-by-step Polarion ALM administration instructions referencing its knowledge base. Sometimes it's the simple techniques. Polarion is straightforward, grounded ,and ready to be used. Check it out! ​
Begin your journey Building 🏗️, Shipping 🚢 and Sharing 🚀
We're offering a limited-time 20% discount for Cohort 8 until this Friday, August 1! Learn to build production-ready LLM applications: RAG and agentic RAG, deep research systems, synthetic data generation and evaluation, application deployment best practices, and more!