DataFramed #63 The Past and Present of Data Science
May 31, 2021
In this discussion, Sergey Fogelson, VP of Data Science and Modeling at ViacomCBS, shares insights from his extensive career. He highlights the dramatic evolution of data science over the last decade, transitioning from outdated systems to modern frameworks. Sergey addresses the ongoing challenges, notably the need for organizational support and effective communication for data initiatives. He underscores the importance of democratizing data access, teaching tools like SQL, and fostering a culture of continuous learning among data professionals.
AI Snips
Chapters
Transcript
Episode notes
Early Projects
- Sergey Fogelson's early data science projects included classifying IP addresses and building device graphs.
- These projects at a digital advertising startup involved large-scale data analysis challenges.
Data Science Evolution
- Data science has evolved from Hadoop to Spark for data processing.
- Data warehousing has become cheaper, enabling structured queries on larger datasets.
Orchestration Revolution
- Orchestration and data pipelining frameworks like Airflow have revolutionized data science workflows.
- These tools automate data jobs and increase workflow velocity.