Booz Allen Hamilton Colloquium: Om Deshmukh, Envestnet

Friday, February 24, 2023
3:30 p.m.
Jeong H. Kim Engineering Building, Room 1110
Darcy Long
301 405 3114

Speaker: Om Deshmukh, Envestnet

Title: "Building At-Scale Data-Driven Products"

Abstract: There is plenty of literature on how to train machine learning models and how to draw inferences from such models. But there is a lack of publicly accessible know-how on how to build real-world data-driven products that cater at big-data scale. In this talk, Om will discuss the framework that he and his team at Envestnet have built and fine-tuned over the years for developing and deploying data-driven products.

The framework will be discussed in the context of Transaction Data Enrichment (TDE): an Envestnet product that processes millions of financial transactions on a daily basis. TDE analyses a given financial transaction to identify various attributes hidden in the transaction description. The TDE system has to solve for several unique data problems: (a) Widely diverse data: Financial transactions have about 50x more words than natural English does, (b) Limited context: An average financial transaction has only 4.5 words, (c) Data paucity: A substantial proportion of the words occur only once when compared to a year’s data.

TDE has to satisfy two additional practical constraints to be a viable product: (a) The compute cost of training and deploying TDE models has to be within a pre-defined compute budget, and (b) TDE has to cater to SLAs of certain minimum prediction accuracies.

The talk will provide under-the-hood details of the various machine learning models that were trained, how these models interact with each other, the tech stack developed for efficient daily processing of millions of transactions and the systematic approach to ingest human subject matter expert's inputs (i.e., ground truth) into a real-world system.

Bio: Om is one of us: He graduated from the ECE department in 2006. His doctoral thesis, under the guidance of prof. Espy-Wilson, who he reveres as his first Guru, was on combining Acoustic Phonetics domain expertise and Machine Learning techniques for enhancing noisy speech signals.

Om is currently the Head of Data Sciences and Head of Data Products at Envestnet: Envestnet’s vision is to make Intelligent Financial Life a reality for everyone. At Envestnet, Om and team drive foundational data science initiatives to mine actionable insights from the peta-bytes of data that flow through their network. His team is also responsible for creating end-to-end data-products and delivering those to the clients.

Om spent close to a decade in IBM Research and Xerox Research driving various Machine Learning initiatives for global technical research, technology strategy and startup partnerships.

His team at IBM won the Govt of India’s National Award for the Best Accessible Technology in 2008. In 2017, he was recognized as one of the top 10 data scientists in India. In 2019, his team at Envestnet was recognized as one of the top 10 data science teams to work for in India.

Om has 50+ international publications and 50+ patents (filed/granted).


Audience: Public  Graduate  Undergraduate  Faculty 

remind we with google calendar


March 2023

26 27 28 1 2 3 4
5 6 7 8 9 10 11
12 13 14 15 16 17 18
19 20 21 22 23 24 25
26 27 28 29 30 31 1
Submit an Event