Generative AI for production workloads & precise market forecasting

Data Science Initiative is proud to bring you an in-person event on LLMs, MLOps, and ML Forecasting in collaboration with our friends at GfK. Join us on the 5th of October and discover how to move from experimentation to production using MLOps methods for precise market forecasting and Large Language Models.

To secure your place, please register at the following link: REGISTRATION LINK

Speakers:

Peter Grimshaw, Machine Learning Engineer, GfK and Yanshan Shi, Senior Machine Learning Engineer, GfK

“GfK Newron Predict Forecasting Machine Learning Models from research to production”

Forecasting has been a key interest for our client. By using GfK POS data, together with events, and some financial indices, the Newron platform predict forecasting team would be able to provide forecasting model, which gives the trend on a market in the next half year. The clients would get benefits using the forecasting results, to make business decisions, such as inventory control, sales events impact analysis, the competitor analysis etc.

MLOps concepts are vital for companies wanting to deliver Machine Learning at scale with scientific accuracy. GfK has been adopting and implementing best practices from the new field of MLOps to ensure that:

• Experimentation is robust.
• We enable rapid transition from research to production.
• We enable data scientists and engineers to work with the same environment.

In this talk we will outline how we are ensuring our forecasting is underpinned by best practices to enable more and faster research and ultimately better models providing valuable insights to our client

Peter Krajcik, Principal Software Engineer, GfK

“Using generative AI to access your data the safe way.”

A possible solution to some of the major problems when providing a LLM powered interface to your data.

• How to prevent data leakage
• How to deal with prompt hacking
• How to deal with the limited context window

Fatos Ismali, Senior Data & AI Solutions Architect, Microsoft

“From prototyping to production - deploying enterprise-level LLM architectures”

Learn how to successfully transition from prototyping to deploying enterprise-level LLM architectures. This talk is designed for data scientists, machine learning engineers, CTOs, and anyone interested in implementing LLMs in an enterprise environment. How to prevent data leakage

• Navigate the current LLM landscape and its relevance in enterprise applications.
• Identify and mitigate common pitfalls during the prototyping phase to ensure a smooth transition to production.
• Implement best practices in scalability, security, and deployment to create a robust, enterprise-level LLM architecture.