At the meetup, Machine Learning evangelists from the Squadex team shared with the attendees (Stanford academia) the following insights:

  1. Successful ML projects always have well-architected data platforms, with the right setup of the local development and cloud infrastructure in place.
  2. AWS ML Ecosystem is a great foundation of your enterprise-wide ML platform, but it has limitations and requires that you build an extra layer on top.
  3. ML Reproducibility and ML lifecycle management is a wild space, with no off-the-shelf solution available yet.

Stanford, CA, Nov. 15, 2018Squadex, a Cloud Transformation Consultancy from Palo Alto, is honored to have delivered their first machine learning meetup on the campus of Stanford University on November 7, 2018.

The Squadex team provided insight into Machine Learning, Data Engineering, and DevOps best practices, with focus on specific ML tools used in gear with AWS products and open-source tools, to show how to build comprehensive cloud solutions.

Topic:
“Tooling for Machine Learning: AWS Products, Open Source Tools, and DevOps Practices”

The keynote was delivered by three speakers:

Stepan Pushkarev, CTO at Squadex

Rinat Gareev, Machine Learning Engineer at Squadex

Iskandar Sitdikov, Machine Learning Engineer at Squadex

The meetup was supported by a special guest — Chaitanya Hazarey, ML Solutions Architect at AWS, who helped address specifics of running ML workloads on AWS cloud.

The speakers walked the attendees through the tooling that assists ML engineers to complete routine tasks (e.g.data ingestion, feature engineering, labeling, parameters tuning, retraining), explored automation do’s and don’ts of using AWS and open-source tools, and shared insights on the specifics of setting up the right local development and cloud infrastructure to achieve the ML lifecycle efficiency.

Specifically, the keynote covered the following areas:

  • ML 101
  • Data pipelines
  • Modeling and training
  • Deployment

As a starting point, the speakers explained how ML works and how the data gets collected, analyzed, and managed in the ML ecosystem. They provided the specifics of using concrete AWS ML tools and moved forward to their limitations. The speakers laid out a few facts about ML model maintenance and monitoring to dig deeper into ML lifecycle orchestration. Finally, they showed a demo on using KubeFlow and compared it to MLFlow.

The attendees walked away with insights and tips on how to use AWS products, open-sources tools, and DevOps practices to set up the local development and cloud infrastructure, and manage the data science life cycle, from research to production. They also recognized why well-architected data platforms are the requirement for any successful ML-powered project and why knowing the limitations of AWS ML ecosystem is important.

The Machine Learning meetup was organized in collaboration with the Technology Training department of Stanford University.

About Squadex
Squadex is a technology consulting and engineering company enabling software delivery automation and data-driven decision-making through DevOps, Big Data, and Machine Learning practices. Squadex assists organizations in improving their IT infrastructure and processes to achieve greater business efficiency and increase ROI.

About Stanford
Stanford is a private research university located in California’s Bay Area, one of the most intellectually dynamic and culturally diverse areas of the nation. Stanford is known for its entrepreneurial character, drawn from its relationship to Silicon Valley, and it is one of the world’s top universities and leading teaching & research institutions. Stanford consists of 40 academic departments and four professional schools. 83 Nobel laureates, 27 Turing Award laureates, and 8 Fields Medalists have been affiliated with Stanford as students, alumni, faculty or staff. Stanford ranks first by the number of alumni-founded startups that raise over $1 million.