--- title: "Why have a Ferrari if you're not going to drive it?" newsletter: "MLOps Community" date: 2024-10-03 source: https://aaif.live/newsletters/mlopscommunity/2024-10-03-why-have-a-ferrari-if-you-re-not-going-to-drive-it --- # Why have a Ferrari if you're not going to drive it? *Plus, the future of feature stores, boosting LLMs mini summit, hybrid RAG reading group, and hidden gems.* *MLOps Community — Agentic AI Foundation, 2024-10-03* https://go.mlops.community/pmyv5c ## Reinvent Yourself and Be Curious // Stefano Bosisio // MLOps Podcast #264 They say mistakes are the best teacher. I’d argue other people’s mistakes are better. The chat covered key topics on transitioning from academia to industry, building ML platforms, and tackling MLOps challenges. Stefano shared his journey from a PhD in computational chemistry to fintech, where he developed ML platforms. His "Ferrari" platform went unused due to missed opportunities for demos and training. Now, with all this hard-won experience, Stefano has built a course to help you ask the right questions when building ML platforms. After this chat—and checking out his course [https://go.mlops.community/xzi4zi]—I’d say Stefano is the best teacher. Video [https://go.mlops.community/lss1ml] || Spotify [https://go.mlops.community/7eumx8] || Apple [https://go.mlops.community/rrnumf] [https://go.mlops.community/rrnumf](https://go.mlops.community/rrnumf) [https://go.mlops.community/lss1ml](https://go.mlops.community/lss1ml) [https://go.mlops.community/7eumx8](https://go.mlops.community/7eumx8) ## Monitor & Evaluate LLMs: Meet Opik by Comet https://go.mlops.community/s4s98j Opik is an open source end-to-end LLM Evaluation Platform that helps developers track their LLM prompts and responses during both development and production. Users can define and run evaluations to test their LLMs apps before deployment to check for hallucinations and accuracy. Opik has native integrations with tools like OpenAI, Langchain, LlamaIndex & more -- meaning it only takes a couple lines of code to add it to your current workflows. Learn more about Opik here [https://go.mlops.community/s4s98j], or explore its GitHub repository here [https://go.mlops.community/1xge29]. ## Unpacking 3 Types of Feature Stores // Simba Khadder // MLOps Podcast #265 This is a sweet episode – and not just because of the analogy of Coke flavors to explain how embeddings in a recommender system work. Simba talks about how feature stores have evolved from basic storage to handling complex production data pipelines. He breaks down how embeddings fit into both feature and vector stores, sharing examples from recommender engines. He also gets into the growing role of LLMs in MLOps, especially when it comes to data personalization. Simba doesn't shy away from the challenges of combining context retrieval with embedding storage, and he sees a bright future with more integration. Have a listen and a smile! Video [https://go.mlops.community/ylq3py] || Spotify [https://go.mlops.community/dn68a1] || Apple [https://go.mlops.community/a2v561] [https://go.mlops.community/a2v561](https://go.mlops.community/a2v561) [https://go.mlops.community/ylq3py](https://go.mlops.community/ylq3py) [https://go.mlops.community/dn68a1](https://go.mlops.community/dn68a1) ## Integrating Knowledge Graphs & Vector RAG for Efficient Information Extraction Integrating Knowledge Graphs & Vector RAG for Efficient Information Extraction Hop in your hybrid car to your hybrid job, to work on hybrid RAG? This session discussed a paper on combining vector retrieval and knowledge graphs to improve accuracy when extracting information from financial documents. We covered the limitations of traditional vector retrieval, which struggles with the hierarchical structure of complex docs, and how knowledge graphs handle relationships better. There was some debate on whether linking vector and graph contexts actually improved things or just added more noise. The faithfulness metric also raised eyebrows, with concerns about longer statements skewing results. Not hybrid, just virtual – watch here [https://go.mlops.community/39txlv]. Register here [https://go.mlops.community/blepmi] for the October 10 session where we'll be covering What is the Role of Small Models in the LLM Era: A Survey [https://go.mlops.community/xwu77b] [watch here](https://go.mlops.community/39txlv) ## Job of the Week [https://go.mlops.community/optn04](https://go.mlops.community/optn04) ## Hidden Gems Interested in partnering with us? Get in touch: partners@mlops.community Thanks for reading. See you in Slack [https://go.mlops.community/iuguh7], YouTube [https://go.mlops.community/mgNff7], and podcast [https://go.mlops.community/ttftlf] land. Oh yeah, and we are also on X [https://go.mlops.community/twitter]. The MLOps Community newsletter is edited by Jessica Rudd [https://go.mlops.community/cetjhj]. --- Source: https://aaif.live/newsletters/mlopscommunity/2024-10-03-why-have-a-ferrari-if-you-re-not-going-to-drive-it