Murari B V

Experience

• 

  Tata Consultancy Services

  May 2021 - Jul 2021

  TCS Research Intern | Physical Sciences Sector

  A deep learning approach towards scalable imaging through scattering media.1. Generation of Bio-Speckles with respect to time with known bio activity of my samples.2. Generation of Speckles of light passing through air medium by performing statistical Monte Carlo Simulation 3. Developed a Deep Neural Network for classification of types of medium with given simulated speckle images.

• 

  ManageArtworks

  Dec 2021 - Jan 2022

  AI Intern
• 

  Indian Institute of Technology, Madras

  Jan 2022 - Jun 2022

  Student Researcher

  Consult Project Under Prof. Usha Mohan, Prof. Srinivasan and Prof. Rahul Marathe. with Viknesh SXYZ (Global Supply Chain Firm) wants to reduce its total cost of ownership by lowering inventory levels. XYZ thinks that, given the scale and complexity of its supply chain, there is a considerable opportunity and need to optimize its inventory in order to achieve a competitive advantage.We optimized the level of safety stock maintained for each material at numerous warehouses across continents. We recommended lowering the level of safety stock by implementing an inventory pooling technique, which would minimize inventory holding costs while also potentially lowering transportation expenses. Show less

• 

  Publicis Sapient

  May 2022 - Jul 2022

  Summer Intern | Data Science

  Dynamic Pricing of products and Optimization of revenue for Online Retail1. Developed an End-End solution with online web application for dynamically pricing various products over a online retail platform2. Various mathematical optimization solutions were considered and final solution was delivered based on product's elasticity and cross product elasticity. 3. In any optimization problem, showing how the final solution is better than anything else is important. and for the client to trust the recommendation given there has to be clear comparison, that was taken care for most of the products.4. Online retail is large, as number of products exponentially increase scaling the optimization becomes extremely important. Solution handled small chunk of product categories which could be customized to handle large number of products. Show less

• 

  Unilever

  Jan 2023 - Jun 2023

  Data Science, Intern

  The forecasting of product sales for a Global FMCG (Unilever), across various categories of products, presents a challenge due to the complexity of the hierarchical structure of both its products and customers. 1. The aim of this study is to improve the accuracy of forecasts at every level of the hierarchy, while ensuring that the forecasts are coherent over the entire hierarchy. 2. To address this issue, we propose the use of reconciliation techniques, commonly used in the literature of hierarchical forecasting.3. Specifically, we recommend the application of dynamic top-down bottom-up (DTDBU)approach which leverages the information from higher hierarchies to generate moreaccurate and consistent forecasts at lower levels4. We also suggest the use of Feed forward Neural Network or CNNs (Kernel Size = 1) for reconciliation which takes care of hierarchy information in its loss function to get much accurate forecasts. Show less

• 

  Publicis Sapient

  Oct 2023 - now

  Senior Associate, Data Science