Scaling up machine-learning (ML), data retrieval and reasoning algorithms from Artificial Intelligence (AI) for massive datasets is a major technical challenge in our time. The scaling process can also have different dimensions: performance, development productivity, number of employees…
In this talk I will showcase how we used to develop Machine learning features at GitHub, the pain points we had and how we changed our infrastructure and way of development in order to productionize multiple ML features in terms of hours/days.
In addition, I will explore with the audience the main factors I consider when scaling ML at medium to big companies.
By the end of the talk you should have an overview and applicable framework on how to help scaling ML processes in your company.
Potential outline for the talk:
- Introduction to ML at GitHub.
- Challenges of running ML at scale. Different dimensions:
- Performance: number of requests
- Development: growing infrastructure, number of ML features
- Organizational: number of employees
- ML ecosystem architecture.
- Improving agility and development on ML features.
- Adopting ML at scale in your company.