Overcoming Barriers to Scale in Machine Learning

Join SigOpt and Slalom as we discuss how to overcome the barriers to building a scalable, repeatable machine learning process.

When: Thursday, November 19 at 6pm PT

There are many barriers to transitioning from developing one-off models to building a scalable, repeatable machine learning process. Overcoming these challenges often requires assembling a combination of data engineering, developer operations, machine learning, software engineering, and domain expertise. Even if you assemble this cross-functional expertise, the messiness of experimentation can make it tough to track model development, manage computing resources, collaborate on machine learning, and reproduce models. 

The best approach combines a defined modeling process, skilled personnel, and machine learning operations software solutions to erode these barriers to scale. In this talk, Slalom draws on their consulting expertise and SigOpt draws on their software expertise to put together a complete picture of how to develop a more scalable approach to machine learning. Attendees will learn how to:

  • Create a framework for defining, interpreting, and evolving their modeling problem
  • Explore value driven modeling and assess risk to ensure long-term applications
  • Define a sustainable, repeatable, rigorous machine learning development process backed by a strong MLOps foundation
  • Track their work with a few lines of code and organize it in an intuitive web dashboard
  • Automate hyperparameter optimization to uncover high-performing model configurations
  • Apply advanced experimentation techniques to understand model behavior more deeply
  • This talk includes a presentation and a tech demo, and attendees will leave with access to a use case, notebook, and API token they can use to recreate key steps in the process.

Through this webinar, attendees will gain access to practical guidance and easy-to-use tooling that are critical for any team looking to scale their machine learning process. 

Robert SiboRob Sibo, Slalom, Practice Director for Data & Analytics Australia

Over 20 years advising and delivering analytics/ML, modern data architecture, visualization and business strategy initiatives. 

Robert has worked across most industries including high technology, retail banking, wealth management and telcom clients in North America, Asia Pacific and in the UK.  Including 4+ years working in Silicon Valley with the tech giants to grow with topics such as scale ML, grapple with analytics in the wake of GDPR..

Robert has authored papers and blogs on topics including: Risk-based ML initiative investing, ML Ops differentiation, Digital workforce, ETL batch architecture, Retail banking data management, Car insurance telematics.



Fay Kallel, SigOpt, Head of Product

Fay Kallel is an entrepreneur, philanthropist and a passionate product leader. She brings two decades of experience incubating and scaling technologies, forming partnerships, and building revenue growth across mobile, advertising and AI industries. 

Fay is currently leading product and design at SigOpt, developing modeling solutions to empower practitioners to build optimal models, collaboratively and at scale. Prior to that Fay led the organization responsible for defining and launching Yahoo! online media buying SaaS platform, video advertising, and mobile sponsored search driving billions in revenues. Earlier in her career, Fay drove product leadership roles at Adobe, Vodafone, Netscape and several early stage startups. Notably Intellisync, where she led Mobile email-push products growing the business to $70MM in 3years. She also led the definition of SyncML standard with the Open Mobile Alliance, resulting in Nokia acquiring Intellisync for $530MM.