�ݺ�ߣshows by User: pingali

�ݺ�ߣshows by User: pingali / http://www.slideshare.net/images/logo.gif �ݺ�ߣshows by User: pingali / Tue, 15 Nov 2022 08:38:50 GMT �ݺ�ߣShare feed for �ݺ�ߣshows by User: pingali Fast Sub-ML Usecase Development.pdf /slideshow/fast-subml-usecase-developmentpdf/254212352 fastsub-mlusecasedevelopment-221115083850-714ce47f
Widespread adoption of machine learning (ML) in industry is still a challenge today due to resource constraints and RoI questions. Production ML approaches today require high skill, rely on large volumes of data, and have long delivery timelines. In this talk, we argue for Sub-ML - a class of ML simpler than traditional ML approaches, often designed to be used in decision support systems, and delivered under tight constraints. Sub-ML, also called as ML-at-reasonable-scale (MLRS) and Analytical ML, covers upto 80% of the ML usecases in an enterprise. Characterized by their speed in realizing business value and support for diverse use cases, Sub-ML applications still require guarantees of correctness, transparency, and auditability in the data transformation process. We draw on our experience in the fin-tech, ed-tech and e-commerce domains to lay out design choices for feature stores to enable Sub-ML, tradeoffs we made including constraining the problem space, bundling capabilities for fast development, and incorporating a data consumption layer.]]>
Widespread adoption of machine learning (ML) in industry is still a challenge today due to resource constraints and RoI questions. Production ML approaches today require high skill, rely on large volumes of data, and have long delivery timelines. In this talk, we argue for Sub-ML - a class of ML simpler than traditional ML approaches, often designed to be used in decision support systems, and delivered under tight constraints. Sub-ML, also called as ML-at-reasonable-scale (MLRS) and Analytical ML, covers upto 80% of the ML usecases in an enterprise. Characterized by their speed in realizing business value and support for diverse use cases, Sub-ML applications still require guarantees of correctness, transparency, and auditability in the data transformation process. We draw on our experience in the fin-tech, ed-tech and e-commerce domains to lay out design choices for feature stores to enable Sub-ML, tradeoffs we made including constraining the problem space, bundling capabilities for fast development, and incorporating a data consumption layer.]]> Tue, 15 Nov 2022 08:38:50 GMT /slideshow/fast-subml-usecase-developmentpdf/254212352 pingali@slideshare.net(pingali) Fast Sub-ML Usecase Development.pdf pingali Widespread adoption of machine learning (ML) in industry is still a challenge today due to resource constraints and RoI questions. Production ML approaches today require high skill, rely on large volumes of data, and have long delivery timelines. In this talk, we argue for Sub-ML - a class of ML simpler than traditional ML approaches, often designed to be used in decision support systems, and delivered under tight constraints. Sub-ML, also called as ML-at-reasonable-scale (MLRS) and Analytical ML, covers upto 80% of the ML usecases in an enterprise. Characterized by their speed in realizing business value and support for diverse use cases, Sub-ML applications still require guarantees of correctness, transparency, and auditability in the data transformation process. We draw on our experience in the fin-tech, ed-tech and e-commerce domains to lay out design choices for feature stores to enable Sub-ML, tradeoffs we made including constraining the problem space, bundling capabilities for fast development, and incorporating a data consumption layer. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/fastsub-mlusecasedevelopment-221115083850-714ce47f-thumbnail.jpg?width=120&height=120&fit=bounds" /> Widespread adoption of machine learning (ML) in industry is still a challenge today due to resource constraints and RoI questions. Production ML approaches today require high skill, rely on large volumes of data, and have long delivery timelines. In this talk, we argue for Sub-ML - a class of ML simpler than traditional ML approaches, often designed to be used in decision support systems, and delivered under tight constraints. Sub-ML, also called as ML-at-reasonable-scale (MLRS) and Analytical ML, covers upto 80% of the ML usecases in an enterprise. Characterized by their speed in realizing business value and support for diverse use cases, Sub-ML applications still require guarantees of correctness, transparency, and auditability in the data transformation process. We draw on our experience in the fin-tech, ed-tech and e-commerce domains to lay out design choices for feature stores to enable Sub-ML, tradeoffs we made including constraining the problem space, bundling capabilities for fast development, and incorporating a data consumption layer.

Fast Sub-ML Usecase Development.pdf from Venkata Pingali

]]> 8 0 https://cdn.slidesharecdn.com/ss_thumbnails/fastsub-mlusecasedevelopment-221115083850-714ce47f-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 Privacy Law Aware ML Data Prep April 2020 /pingali/privacy-law-aware-ml-data-prep-april-2020 privacylaw-awaremldataprepapril2020-200331064451
Talk at ODSC]]>
Talk at ODSC]]> Tue, 31 Mar 2020 06:44:51 GMT /pingali/privacy-law-aware-ml-data-prep-april-2020 pingali@slideshare.net(pingali) Privacy Law Aware ML Data Prep April 2020 pingali Talk at ODSC <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/privacylaw-awaremldataprepapril2020-200331064451-thumbnail.jpg?width=120&height=120&fit=bounds" /> Talk at ODSC

Privacy Law Aware ML Data Prep April 2020 from Venkata Pingali

]]> 180 0 https://cdn.slidesharecdn.com/ss_thumbnails/privacylaw-awaremldataprepapril2020-200331064451-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 Accelerating ML using Production Feature Engineering /slideshow/accelerating-ml-using-production-feature-engineering/177697120 acceleratingmlusingproductionfeatureengineering-190930093311
�ݺ�ߣs from ODSC 2019 Talk]]>
�ݺ�ߣs from ODSC 2019 Talk]]> Mon, 30 Sep 2019 09:33:11 GMT /slideshow/accelerating-ml-using-production-feature-engineering/177697120 pingali@slideshare.net(pingali) Accelerating ML using Production Feature Engineering pingali �ݺ�ߣs from ODSC 2019 Talk <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/acceleratingmlusingproductionfeatureengineering-190930093311-thumbnail.jpg?width=120&height=120&fit=bounds" /> �ݺ�ߣs from ODSC 2019 Talk

Accelerating ML using Production Feature Engineering from Venkata Pingali

]]> 303 3 https://cdn.slidesharecdn.com/ss_thumbnails/acceleratingmlusingproductionfeatureengineering-190930093311-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 Reducing Cost of Production ML: Feature Engineering Case Study /pingali/reducing-cost-of-production-ml-feature-engineering-case-study fifthelephant-featureengineeringcasestudy-190405073414
Early talk given at Fifth Elephant Winter Edition at Mumbai in Jan, 2019]]>
Early talk given at Fifth Elephant Winter Edition at Mumbai in Jan, 2019]]> Fri, 05 Apr 2019 07:34:14 GMT /pingali/reducing-cost-of-production-ml-feature-engineering-case-study pingali@slideshare.net(pingali) Reducing Cost of Production ML: Feature Engineering Case Study pingali Early talk given at Fifth Elephant Winter Edition at Mumbai in Jan, 2019 <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/fifthelephant-featureengineeringcasestudy-190405073414-thumbnail.jpg?width=120&height=120&fit=bounds" /> Early talk given at Fifth Elephant Winter Edition at Mumbai in Jan, 2019

Reducing Cost of Production ML: Feature Engineering Case Study from Venkata Pingali

]]> 438 7 https://cdn.slidesharecdn.com/ss_thumbnails/fifthelephant-featureengineeringcasestudy-190405073414-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 Using dataset versioning in data science /slideshow/using-dataset-versioning-in-data-science/61747070 usingdatasetversioningindatascience-160506140940
Slight variation of the R Data Scientist talk given at the Machine Learning Bangalore Meetup on April 30, 2016]]>
Slight variation of the R Data Scientist talk given at the Machine Learning Bangalore Meetup on April 30, 2016]]> Fri, 06 May 2016 14:09:39 GMT /slideshow/using-dataset-versioning-in-data-science/61747070 pingali@slideshare.net(pingali) Using dataset versioning in data science pingali Slight variation of the R Data Scientist talk given at the Machine Learning Bangalore Meetup on April 30, 2016 <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/usingdatasetversioningindatascience-160506140940-thumbnail.jpg?width=120&height=120&fit=bounds" /> Slight variation of the R Data Scientist talk given at the Machine Learning Bangalore Meetup on April 30, 2016

Using dataset versioning in data science from Venkata Pingali

]]> 613 4 https://cdn.slidesharecdn.com/ss_thumbnails/usingdatasetversioningindatascience-160506140940-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 Analytics Lessons Learnt /slideshow/analytics-lessons-learnt/60370725 analytics-lessonslearnt-160402045518
Presentation to FourthLion (my former employer) staff on some lessons learnt while doing analytics across three domains, and the motivation for automation. Data science will IMO (a) have significant growing pains (b) see evolution similar to those that we saw in software engineering. ]]>
Presentation to FourthLion (my former employer) staff on some lessons learnt while doing analytics across three domains, and the motivation for automation. Data science will IMO (a) have significant growing pains (b) see evolution similar to those that we saw in software engineering. ]]> Sat, 02 Apr 2016 04:55:18 GMT /slideshow/analytics-lessons-learnt/60370725 pingali@slideshare.net(pingali) Analytics Lessons Learnt pingali Presentation to FourthLion (my former employer) staff on some lessons learnt while doing analytics across three domains, and the motivation for automation. Data science will IMO (a) have significant growing pains (b) see evolution similar to those that we saw in software engineering. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/analytics-lessonslearnt-160402045518-thumbnail.jpg?width=120&height=120&fit=bounds" /> Presentation to FourthLion (my former employer) staff on some lessons learnt while doing analytics across three domains, and the motivation for automation. Data science will IMO (a) have significant growing pains (b) see evolution similar to those that we saw in software engineering.

Analytics Lessons Learnt from Venkata Pingali

]]> 244 4 https://cdn.slidesharecdn.com/ss_thumbnails/analytics-lessonslearnt-160402045518-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 R meetup talk scaling data science with dgit /slideshow/r-meetup-talk-scaling-data-science-with-dgit/60056226 rmeetuptalk-scalingdatasciencewithdgit1-160326121537
First talk on the need for dataset versioning, structure of dgit, and demo. This was R data science meetup at Bangalore on March 26]]>
First talk on the need for dataset versioning, structure of dgit, and demo. This was R data science meetup at Bangalore on March 26]]> Sat, 26 Mar 2016 12:15:37 GMT /slideshow/r-meetup-talk-scaling-data-science-with-dgit/60056226 pingali@slideshare.net(pingali) R meetup talk scaling data science with dgit pingali First talk on the need for dataset versioning, structure of dgit, and demo. This was R data science meetup at Bangalore on March 26 <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/rmeetuptalk-scalingdatasciencewithdgit1-160326121537-thumbnail.jpg?width=120&height=120&fit=bounds" /> First talk on the need for dataset versioning, structure of dgit, and demo. This was R data science meetup at Bangalore on March 26

R meetup talk scaling data science with dgit from Venkata Pingali

]]> 646 5 https://cdn.slidesharecdn.com/ss_thumbnails/rmeetuptalk-scalingdatasciencewithdgit1-160326121537-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 https://cdn.slidesharecdn.com/profile-photo-pingali-48x48.jpg?cb=1668501488 I am an academic turned entrepreneur working at the intersection of data and systems. I enjoy solving messy problems that require learning by doing, strong technical understanding, and analytical thinking. At Scribble, I am building tooling to increase pace of organizational learning from data. I am always open to stimulating conversations. Feel free to drop me a note or connect with me. https://cdn.slidesharecdn.com/ss_thumbnails/fastsub-mlusecasedevelopment-221115083850-714ce47f-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/fast-subml-usecase-developmentpdf/254212352 Fast Sub-ML Usecase De... https://cdn.slidesharecdn.com/ss_thumbnails/privacylaw-awaremldataprepapril2020-200331064451-thumbnail.jpg?width=320&height=320&fit=bounds pingali/privacy-law-aware-ml-data-prep-april-2020 Privacy Law Aware ML D... https://cdn.slidesharecdn.com/ss_thumbnails/acceleratingmlusingproductionfeatureengineering-190930093311-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/accelerating-ml-using-production-feature-engineering/177697120 Accelerating ML using ...