Ashish Verma

Experience

• 

  Indian Institute of Technology, Kanpur

  Jun 2011 - Nov 2011

  Research Associate

  District level project entitled “Accelerated Life Testing” - Six month research associate role in IITK. - Type-1 and Type-2 censoring with Lindley distribution and to derive the statistical properties.

• 

  RSG Media

  Sept 2012 - May 2015

  - Developed and implemented mathematical optimization model using FICO’s Xpress solver for plan-optimization and spot-optimization to improve revenue yield for media channels. - Developed a look-a-like model for scoring users, based on their propensity of purchasing a product on a control group and generalized over the whole population.- Developed forecasting model to forecast the impressions over the offsets i.e. the estimated number of viewers, going to watch that particular TV Chanel in given time duration.- R, SQL, Mathematical Optimization, Predictive Modeling

  • Sr. Statistical Analyst

    Apr 2014 - May 2015

  • Statistical Analyst

    Sept 2012 - Mar 2014
• 

  Vserv AudiencePro

  May 2015 - Aug 2017

  - CTR, CVR prediction(scalable response prediction for display advertising), - AdaDelta algorithm to estimate learning rate in gradient decent optimization method.- Ads optimization algo based on Bandit logic(SoftMax, UCB approach)- A/B testing- Python, AWS Framework, Hive, Hadoop System - Developed App Monetization algorithm for supply side platform called VMAX to optimize yield by selecting the best ad partners to bring the highest eCPMs, resulting in maximized yield.- Time Series forecasting

  • Sr. Data Scientist

    May 2016 - Aug 2017

  • Data Scientist

    May 2015 - Apr 2016
• 

  Accenture

  Sept 2017 - Nov 2018

  Data Scientist Specialist

  ML based solution to categorize invoices (~200 millions invoices) and to replace the status-quo operations to completely in automated predictive frame in procurement process.Tools: Python, Vowpal Wabbit, SQL, Linux

• 

  日産自動車株式会社

  Nov 2018 - Jan 2022

  Data Science Manager

  - Global After-sales business: Led the Data Science capabilities for Nissan Global Digital After-sales business vertical.- CRM 360-degree to provide better customer service experience.- Finance Vertical: Machine learning method to priorities collection treatments and to increase funds collection for delinquent accounts of the auto loan portfolio.

• 

  Ericsson Software

  Jan 2022 - now

  Sr. Data Scientist

  Utilizing the data science approach to optimize the Energy Consumption at RRU Level: - Created a system capable of identifying inefficient RRUs (Radio) based on energy consumption efficiency and other parameters in telecom data within a specific region. - Designed a Reinforcement-Learning based algorithm to optimize energy consumption at the RRU level.Developed an automated process to find the network performance degradation and determine the causal inference for degradations in data centers. - Anomaly detection in the univariate and multivariate time series KPIs data. - Causal inference using PCMCI that estimates causal effects on a causal graph. - the temporal causal discovery framework, TCDF uses attention-based convolutional neural networks to learn a causal graph structure by discovering causal relationships in observational time series data. Show less