Learnings from LLMs and MLOps in Production

We’ll be meeting at the Pendulum office in Smith Tower. This time sponsored by [Zilliz](https://www.linkedin.com/company/zilliz/mycompany/), [Iterative](https://www.linkedin.com/company/iterative-ai/), and [Pendulum](https://www.pendulumfn.com/).

PLEASE FILL OUT THIS FORM SO SMITH TOWER WILL LET YOU IN! https://forms.gle/3CaRZuquszz7MsNC8 (also provides opt-in/opt-out from receiving emails from the sponsors)

Zilliz is the company behind the world’s most popular open source vector database, [Milvus](https://github.com/milvus-io/milvus), with just over 23k stars and over 4M downloads. Milvus is built to scale with a separation of concerns and a distributed system architecture.

Iterative is the company behind DVC. DVC standardizes real-world machine learning workflows to lower the cost of building and running end-to-end workflows from months/weeks to days/hours. Manage your data, experiments, and models in one place, and connect it all with your code using best practices borrowed from software and adapted to meet the challenges of machine learning.

Pendulum’s cutting-edge AI and ML make sense of vast data streams, monitoring close-range threats, such as violent or antisemitic posts about your brand or executives, and over-the-horizon risks, like boycotts, geopolitical shifts, ESG, or taxation, and uncovers how they are impacting your business.

**What we’ll do:**
We’ll get a series of three talks about large language models in production.

5:30-6:00: Doors open for arrival
6:00-6:30: Implementing MLOps in AI-Driven Organizations – [Diego Kiedanski](https://www.linkedin.com/in/diego-kiedanski/) from Tryolabs
6:35-7:05: Pendulum, TBA
7:10-7:40: Retrieval Augmented Generation and Its Flavors – [Yujian Tang](https://www.linkedin.com/in/yujiantang/)
7:40-8:30: Networking

**Who Should attend:**
People who want to learn more about what people who are working with LLMs are experiencing, what problems they’re running into, what

**When:**
Thursday, October 19th, 2023
5:30pm – 8:30pm.

**Where is it:**
Smith Tower

**Speaker Bios**

**Diego:**
Diego Kiedanski is a highly accomplished AI Solution Architect at Tryolabs, where he leverages his extensive expertise in computer engineering and artificial intelligence to create innovative solutions for clients. With a strong academic background that includes a degree in Computer Engineering from Uruguay, a Ph.D. in game theory from IP Paris, and postdoctoral research in distributed machine learning systems at Yale University, Diego, is well-equipped to tackle complex challenges. As a dedicated educator, he also shares his knowledge as a professor, teaching courses on AI and its applications in the business world. Diego’s unique blend of technical acumen and passion for education makes him an invaluable asset to both his clients and students.

**Yujian:**
Yujian started off his software career interning at IBM after winning ACSL in 2012 and 2013. From there, he expanded into machine learning and big data research, publishing to IEEE Big Data as an undergrad. He then led and built two multi-million dollar generating projects for Amazon Retail’s AutoML system. Following his stint at Amazon, Yujian has worked in startups, particularly focusing on the AI ecosystem. His educational software blogs, videos, and talks, have reached over a quarter million software developers.

**Speaker Abstracts**

**Diego:**
As AI continues to reshape industries, the challenge lies in efficiently deploying and scaling these solutions. MLOps, the fusion of Machine Learning and operational practices, offers a strategic guide for AI-driven enterprises. Drawing from over a decade of experience, we’ll share industry insights on applying MLOps, contrasting the theory with its practical application. In this talk, we’ll discuss the challenges we’ve faced, the lessons learned, and the issues we are yet to resolve. We’ll also highlight MLOps’ role in translating AI’s potential into business results and discuss the critical human element, touching on both technical and organizational changes required for successful implementation.

**Yujian:**
Retrieval Augmented Generation is a huge use case for large language models. Yujian’s talk addresses three primary questions. Why do you care about RAG? How does RAG work? What are some main concerns with RAG?

**Hosted by:**

* [Sage Elliott](https://www.linkedin.com/in/sageelliott/) at [WhyLabs](https://whylabs.ai/)
* [Yujian Tang](https://www.linkedin.com/in/yujiantang/) at [Zilliz](https://zilliz.com/)
* [Jenifer De Figueiredo](https://www.linkedin.com/in/jenifer-de-figueiredo?miniProfileUrn=urn%3Ali%3Afs_miniProfile%3AACoAAAKTgzcBwZtHJezWCtDkWotEmJUSVZRVomk&lipi=urn%3Ali%3Apage%3Ad_flagship3_search_srp_all%3BhLovO4JUTVSdRezmevLDqw%3D%3D) at [IterativeAI](https://iterative.ai/)
* Join the #seattle channel on the MLOps slack to stay connected and ask questions! [https://go.mlops.community/slack](https://go.mlops.community/slack)