--- title: "DS = Data Scientist = Dynamic Skills" newsletter: "MLOps Community" date: 2024-07-11 source: https://aaif.live/newsletters/mlopscommunity/2024-07-11-ds-data-scientist-dynamic-skills --- # DS = Data Scientist = Dynamic Skills *Plus, eliminate the hate, Meta musings, multimodal domain fusions, finding code, mocking LLMs, and hidden gems.* *MLOps Community — Agentic AI Foundation, 2024-07-11* ## All Data Scientists Should Learn Software Engineering Principles // Catherine Nelson // Podcast #245 With data scientist job descriptions being so frustratingly vague, it’s no surprise to see the transition to ML roles. That’s why this was a useful chat about the key skills data scientists need to thrive in production environments, including version control, API understanding, security practices, and code quality. We chatted about the challenges of transitioning from Jupyter notebooks to scalable systems, touching on topics like CI/CD, testing, and the debate around data scientists learning Kubernetes. It was also interesting to talk about how model evaluation may reunite diverging data science skill sets. Here’s an easy job description: click below and have a listen. Video [https://home.mlops.community/home/videos/why-all-data-scientists-should-learn-software-engineering-principles] || Spotify [https://open.spotify.com/episode/1kXMajL4HiWKddGiK5Fmbw?si=EQef7a4tTSSJw06G-hR6GQ] || Apple [https://podcasts.apple.com/us/podcast/all-data-scientists-should-learn-software-engineering/id1505372978?i=1000661253607] [https://podcasts.apple.com/us/podcast/all-data-scientists-should-learn-software-engineering/id1505372978?i=1000661253607](https://podcasts.apple.com/us/podcast/all-data-scientists-should-learn-software-engineering/id1505372978?i=1000661253607) [https://home.mlops.community/home/videos/why-all-data-scientists-should-learn-software-engineering-principles](https://home.mlops.community/home/videos/why-all-data-scientists-should-learn-software-engineering-principles) [https://open.spotify.com/episode/1kXMajL4HiWKddGiK5Fmbw?si=EQef7a4tTSSJw06G-hR6GQ](https://open.spotify.com/episode/1kXMajL4HiWKddGiK5Fmbw?si=EQef7a4tTSSJw06G-hR6GQ) ## AI For Good - Detecting Harmful Content at Scale // Matar Haller // MLOps Podcast #245 Haters gonna hate. Which is why I really appreciate the work Matar and her team do. She talked about combating hate speech, child abuse, and other harmful content using AI, including transformer and ensemble models. If there’s any area you want to highlight the importance of data and continuous model reviewing, this is it. She talked about the continuous auditing and retraining of AI models to maintain accuracy and adapt to new forms of harmful content. She also discussed how they prioritize the psychological well-being of data labelers and how AI will take some of that burden off content moderators. No hate, just love for this episode. Video [https://home.mlops.community/home/videos/ai-for-good-detecting-harmful-content-at-scale] || Spotify [https://open.spotify.com/episode/7om8EFHAQn5r85E9nAFQyJ?si=Dqe6jZQVRGabIHimB0z3cw] || Apple [https://podcasts.apple.com/us/podcast/ai-for-good-detecting-harmful-content-at-scale-matar/id1505372978?i=1000661687231] [https://podcasts.apple.com/us/podcast/ai-for-good-detecting-harmful-content-at-scale-matar/id1505372978?i=1000661687231](https://podcasts.apple.com/us/podcast/ai-for-good-detecting-harmful-content-at-scale-matar/id1505372978?i=1000661687231) [https://home.mlops.community/home/videos/ai-for-good-detecting-harmful-content-at-scale](https://home.mlops.community/home/videos/ai-for-good-detecting-harmful-content-at-scale) [https://open.spotify.com/episode/7om8EFHAQn5r85E9nAFQyJ?si=Dqe6jZQVRGabIHimB0z3cw](https://open.spotify.com/episode/7om8EFHAQn5r85E9nAFQyJ?si=Dqe6jZQVRGabIHimB0z3cw) ## Meta GenAI Infra Blog Review // Special MLOps Podcast If I refer to this intro in the intro, that’s pretty meta. What’s even more Meta is their three blog posts I chat about in this episode. There’s some observations, and a look at the design pillars and what can be learned from creating LLaMA, like scalability, and optimal GPU connectivity. Then there’s a look at the infrastructure stack and maintenance of a million operations per day, ensuring efficient AI training processes, and the use of open-source projects like Grand Teton for computing needs. Now I just need to do a blog post about the podcast about blog posts—be mega meta. Spotify [https://open.spotify.com/episode/5EJZZ0eMVmRRTAQAhqge4q?si=pUYpqJkBRNGKyNuAmgeKOA] || Apple [https://podcasts.apple.com/us/podcast/meta-genai-infra-blog-review-special-mlops-podcast/id1505372978?i=1000661057104] [https://podcasts.apple.com/us/podcast/meta-genai-infra-blog-review-special-mlops-podcast/id1505372978?i=1000661057104](https://podcasts.apple.com/us/podcast/meta-genai-infra-blog-review-special-mlops-podcast/id1505372978?i=1000661057104) [https://open.spotify.com/episode/5EJZZ0eMVmRRTAQAhqge4q?si=pUYpqJkBRNGKyNuAmgeKOA](https://open.spotify.com/episode/5EJZZ0eMVmRRTAQAhqge4q?si=pUYpqJkBRNGKyNuAmgeKOA) [https://open.spotify.com/episode/5EJZZ0eMVmRRTAQAhqge4q?si=pUYpqJkBRNGKyNuAmgeKOA](https://open.spotify.com/episode/5EJZZ0eMVmRRTAQAhqge4q?si=pUYpqJkBRNGKyNuAmgeKOA) ## Job of the Week [https://boards.greenhouse.io/gitlab/jobs/7522926002](https://boards.greenhouse.io/gitlab/jobs/7522926002) ## Multimodal is Here Or is it? // Yi Ding // MLOps Community IRL Meetup #85 Silicon Valley ## MLOps Community IRL Meetup Domain fusion – not the latest trendy place to try sushi burritos. Instead, it’s something essential for advancing LLMs, especially multimodal ones. This talk explains how it enables systems to manage and interpret language complexities more holistically, leading to more accurate and nuanced understanding and generation. It covers the potential and challenges of multimodal models, highlighting their lag behind text-based models and the opportunity for growth. The importance of community engagement in advancing these technologies is also emphasized. A talk with plenty of takeaways. Watch it here [https://home.mlops.community/home/videos/multimodal-is-here-or-is-it] [Watch it here](https://home.mlops.community/home/videos/multimodal-is-here-or-is-it) ## Find Your Code! Scaling a LlamaIndex and Qdrant Application with Google Kubernetes Engine Matching socks; where you parked your car; the end of the roll of tape—some things are just hard to find. Helpfully, this blog will help you find that bit of code you wrote months ago. It outlines setting up a GKE cluster, deploying Qdrant and LlamaIndex containers that will allow you to interact with your GitHub repositories, making it easier than ever to find forgotten code snippets. We’ll deploy the application on Google Kubernetes Engine (GKE) with Docker and FastAPI and provide an intuitive Streamlit UI for sending queries. The link’s just at the top, in case you had trouble finding it. With thanks to Benito Martin for their contribution. Effective Practices for Mocking LLM Responses During the Software Development Lifecycle [https://home.mlops.community/home/blogs/effective-practices-for-mocking-llm-responses-during-the-software-development-lifecycle] Mock (yeah) ing (yeah) LLM (yeah) yeah (yeah). This blog takes a look at effective practices for mocking LLM responses. It addresses the challenges of LLM variability and resource intensity, highlighting when and how to mock LLM responses. Key sections include the limitations of traditional testing methods, the benefits of mocking in specific scenarios (like saving costs and unblocking developers), and detailed methods for mocking responses using frameworks and libraries. The article also emphasizes the integration of mocking into different test levels, despite the potential for random test failures, to ensure reliable application performance in production. Hopefully the song will be out of your head by the time you’ve read it. With thanks to Vuong Ngo for their contribution. [Find Your Code! Scaling a LlamaIndex and Qdrant Application with Google Kubernetes Engine](https://home.mlops.community/home/blogs/find-your-code-scaling-a-llamaindex-and-qdrant-application-with-google-kubernetes-engine) ## Hidden Gems ## IRL Meetups San Francisco [https://lu.ma/tmtod6ia] - July 16 Lisbon [https://lu.ma/7sowmead] - July 18 New York [https://lu.ma/6v4jcdnt] -July 18 Denver [https://mlops.community/event/pizza-pints-expert-speaker-intro-to-machine-learning-toolkits-with-kubeflow/] - July 30 Bucharest [https://lu.ma/gbi3flph] - July 31 [San Francisco](https://lu.ma/tmtod6ia) Thanks for reading. See you in Slack [https://go.mlops.community/slack], YouTube [https://www.youtube.com/channel/UCG6qpjVnBTTT8wLGBygANOQ?view_as=subscriber], and podcast [https://home.mlops.community/public/content/] land. Oh yeah, and we are also on X [https://twitter.com/mlopscommunity]. The MLOps Community newsletter is edited by Jessica Rudd [https://www.linkedin.com/in/jmrudd/]. --- Source: https://aaif.live/newsletters/mlopscommunity/2024-07-11-ds-data-scientist-dynamic-skills