Applied Digital, Data & AI

Chairpersons welcome and opening remarks.

• The kinds of experiences that harness the power of AI: conversational, multimodal, and XR.
• The senses: understanding why people need experiences that use AI.
• The key capabilities that can be unlocked by AI-driven user interfaces.
• The new opportunities: what are the key products where AI-driven experiences are valuable.

Companies around the world are spending billions to harness the power of this data to make better and faster business decisions. However, many modern data initiatives are not able to produce the intended value as companies are struggling to operationalize modern data initiatives.. In fact, some analysts estimate that only 15% of modern data initiatives will make it to production! Join this session to learn why DataOps is becoming an important practice in data engineering to not only get data projects into production faster but also run them at scale.

ChatGPT’s triumph over stone-age AI, a tale yet untold,

Boldly snatching the limelight, its brilliance unfold.

BringingAI within everyday’s reach, its power to unfold,

LLM and ChatGPT, energy problems we must behold.

Boundless potential, challenges laid bare,

Enterprises embracing, future’s suspense in the air.

Untapped use-cases, waiting to befound, verified, shared,

New trials await, AI’s potential dared.

Unlock LLM use-cases, find your way,

Embrace the impact, seize the day.

• The rapid rollout of the augmented workforce due to labour shortages – what does this currently look like across enterprises?  
• Aiding employees’ skills and capabilities with AI, ML and intelligent automation to help build a digitally enabled and efficient workforce.   
• Discussing the idea that an augmented workforce can boost employee satisfaction due to automation of repetitive tasks and more time for innovation.  
• It takes two – tackling the idea that humans are being replaced by technologies.  

• The challenges of customizing large foundational models (e.g., ChatGPT) for an enterprise.
• How to train ML modes on enterprise data that is private, sensitive, and siloed?
• How to leverage collaborative AI to reduce the costs of AI development and deployment?
• An overview of FedML AI ecosystem to empower large-scale collaborations for customization of AI models.

• Your EtLT Strategy – When best to apply ETL, and when to apply ELT. 
• Building your toolset to serve your business needs. 
• How to define what’s important when you build your data architecture. 

Stitch Fix is a client-centric styling service that makes personalized styling accessible to everyone. Combining the human touch of expert stylists with the precision of advanced AI models allows Stitch Fix to interpret its billions of unique data points on style and fit and provide 3.6M clients with highly personalized recommendations at scale. This talk will cover how this art and science approach to styling has set Stitch Fix apart from traditional retailers.

Live streamed presentation 

In this talk, we will delve into the significance of feature stores in the era of machine learning that we are entering. Complex features such as embeddings begin to replace certain traditional features. Vector databases are added to the mix. Drawing on my experience working on recommender systems that employed deep learning models that used both traditional features and embeddings, I will showcase how these two worlds combine and where feature stores fit in. By the end of the talk, attendees will gain insights into the importance of feature stores in modern machine learning workflows and how they can be effectively used in a Post-GPT world.  

• How we came to virtual reality.
• The adaptation period and the introduction of virtual reality simulators in our own training center.
• Development of software for simulators.
• Prospects for the development of VR.

Building ML solutions from scratch is hard, requires expert knowledge and lots of low-level code. New declarative ML platforms — pioneered at Uber, Apple and Meta— enable engineers and data scientists to easily build ML-powered applications in just a few lines of code with a simple configuration file.

In this talk, Piero Molino, creator of open-source Ludwig and founder of Predibase, will share how you can use declarative ML frameworks to rapidly build and deploy end-to-end ML pipelines in minutes. He’ll also share how declarative ML can be used to build your own GPT-like application with privately hosted and customized LLMs.

• The current work within the NLP (Natural Language Processing) landscape – examples of real-life use-cases.  
• What industries could benefit from utilising NLP capabilities?  
• The next steps and progressions that are needed to break barriers.

• High granularity scenario-level testing.
• Automated regression testing.
• Systematic model validation.
• Best practices for systematic model testing.

Worldwide organizations are making NLP at the heart of their data science team’s work for various areas of the business. Dedicating time and investment to NLP is catapulting some to far better returns as they see real, actionable insights, improve customer satisfaction, and streamline processes to reduce costs. The scope of NLPs is on the brink and only set to grow as this area of technology grows. 

Large language models (LLMs) have recently landed in the public spotlight for their ability to generate text that is indistinguishable from human-written text. In this talk, I will show how LLMs can be used to generate text for a variety of practical business applications, including generating creative content, translating languages, and answering questions in an informative way. I will also discuss some of the limitations of LLMs and how they can be improved in the future.

View day 2 content here:

• Optimised Data & Analytics
• Transformational AI
• Human-centred Approaches to DTX