The New Frontiers of Machine Learning

Omri Allouche – Head of Research at Gong

In the age of GPT and LLMs, technology companies have the opportunity to revolutionize the way they build products. These advanced models have the potential to greatly speed up the development process, but in order to build products that people actually use and love, a skilled data science team is crucial. In this talk, we will discuss the role of data scientists in this new landscape, explore how they can best make use of GPT and LLMs to drive innovation, and discuss how to best organize data science teams to stay ahead of the curve in the fast-paced world of data science and technology.

Jigyasa Grover – Senior Machine Learning Engineer at Twitter

In the contemporary world of machine learning algorithms – “data is the new oil”. For the state-of-the-art ML algorithms to work their magic it’s important to lay a strong foundation with access to relevant data. Volumes of crude data are available on the web nowadays, and all we need are the skills to identify and extract meaningful datasets. This talk aims to present the power of the most fundamental aspect of Machine Learning – Dataset Curation, which often does not get its due limelight. It will also walk the audience through the process of constructing good quality datasets as done in formal settings with a simple hands-on Pythonic example. The goal is to institute the importance of data, especially in its worthy format, and the spell it casts on fabricating smart learning algorithms.

Ameya Diwan – Ethical AI Analyst at Rocket Central

Brittney Monroe – Sr. Engineer at Carvana
Christina Taylor – Sr. Engineer II at Carvana

Previously, the NLP model training was a manual process. These steps included piecemeal jobs spread across multiple GCP projects with various timing/scheduling. Airflow enables us to automate the entire process on a schedule or on-demand with little to no human intervention. We can now break down a monolithic job into several dependent components. This prevents full job failure, allows us to reprocess independently, and train models faster in parallel. We further implemented GKEStartPodOperator to isolate dependencies and spin up customizable resources as needed, as well as incorporated Great Expectations for data quality checks.

Serg Masís – Climate & Agronomic Data Scientist at Syngenta

Trust is mission-critical for any technology, so if AI solutions are to supplant software, AI must reach the reliability standards currently expected from software. For that to happen, a new field of MLOps engineering has branched off from the DevOps. Also, Explainable AI (XAI) will be more widely adopted since it provides the toolset to interpret machine learning predictions and scrutinize metrics. To ensure increased reliability, and robustness new roles for Machine Learning Quality Assurance will appear likely within DevOps, SecOps, and MLOps teams, but also the roles of data scientist and Machine Learning engineer will evolve.

We will examine examples and discuss how they can revolutionize the way we train, but most importantly, evaluate and deploy machine learning models with examples from the agribusiness industry and the digital agronomy field.

Ali Vanderveld – Senior Staff Scientist at Wayfair

Recommender systems are now ubiquitous in e-commerce. At Wayfair, we use a collection of machine learning models to predict which content and which products to show our customers at every stage in their shopping journey. Models for recommendations are typically trained to optimize for near-term measures of customer satisfaction, such as conversion. However, there are other business objectives that can appear to be in conflict with this idealized ranking. For example, some products might have higher margins, and surfacing such products higher could boost long-term profit. However, such boosts will likely come with a decrease in short term conversion. Thus we need to find ways to balance these competing objectives if we want to make our product recommendations “profit aware.” In this talk I will give an overview of this problem and how we have been addressing it at Wayfair.

Wei Li – VP and GM, Artificial Intelligence and Analytics (AIA) at Intel Corporation

We have a great opportunity to democratize AI. The challenge is that AI is very compute intensive, which often requires additional specialized AI accelerators. However, CPU is ubiquitous. With the recent built-in acceleration for AI in mainstream CPU, we have a breakthrough in making AI widely available. In this talk, we will show an amazing application, Stable Diffusion, on Intel’s latest CPU, and take a peek inside the software and hardware that make this possible.

Iswarya Murali – Principal Data Scientist at Microsoft
Madonna M Dodd – Principal Data Engineer at Microsoft

“Data is food for AI” – Andew Ng. According to Forbes, Data Scientists spend approximately 80% of their time collecting, cleaning and organizing data before even starting the exploratory analysis phase. This particular task could be mundane and often frustrating but is critical. Producing data that is consistent, reliable, available, and of high quality is crucial to the success of Machine Learning projects. The consequences of not investing enough into data and analytics engineering could result in feeding incomplete (or worse, incorrect) training data into models. In this talk, we’ll discuss the process of building a “data muscle” and how to incorporate engineering principles such as Scalability, User Privacy, Maintainability, Code Modularity, and Redundancy in order to create a well-functioning and impactful data team. 

Sandhya Gopalan – AI & MLOps Practice Head at EY Global Delivery Services

Mikiko Bazeley – Head of MLOps at Featureform

Angelica Lo Duca – Researcher at IIT-CNR

Data storytelling is the art of using data to tell a story, communicate insights, and invite an audience to action. Comet is an experimentation platform that enables you to track and monitor your machine learning experiments during almost the project life cycle. In Comet, data storytelling can be accomplished through panels and reports. Panels are customizable visualizations that allow users to explore and analyze data in real time. On the other hand, reports present data in a structured format, often including charts, graphs, and tables. This talk describes how to combine reports and panels in Comet to communicate the story behind data to stakeholders, making it easier to identify trends, discover insights, and drive decision-making.

Andy McMahon – Head of MLOps at NatWest Group

Ronen Dar – CTO at Run.ai

Ozlem Yaylaci – Principal Data Scientist at Oshkosh Corporation

An increasing number of organizations are creating data science teams to recognize the potential of their data to boost performance. Since data science is a relatively new field, however, most companies have a hard time integrating their data science teams to the larger organization, and data scientists can find themselves in a lonely corner trying to manage unrealistically high expectations, without proper guidance from the business.

In this talk, we will explore best practices for creating a diverse and cross-functional data science team that can integrate to the business and harness their full potential, to the benefit of the organization.

Hugo Bowne – Head of Developer Relations at Metaflow/Outerbound

Doug Blank – Head of Research at Comet
Caleb Kaiser – Software Engineer at Comet