Daipayan Sinha Roy

Experience

• 

  Kantar

  Aug 2017 - Aug 2018

  Analytics Executive

  Created structured databases including some statistical outputs out of unstructured raw SPSS datasets based on business logics required by internal market researchers.

• 

  SMART CONSULTING SOLUTIONS PTE LTD

  Aug 2018 - Jun 2019

  Statistician - Radix Analytics

  Developed a web application using R shiny to predict price per square feet of real estate projects. The backend model was developed by analyzing the historical data of real estate projects including their launch prices, amenities, specifications etc. Used word embedding techniques like word2vec and machine learning models to come up with a final predictive model for price estimation.Predicted the behaviour scores of the customers of a banking client who applied for personal loans using an in house application namely scorebuilder developed using R shiny. The application includes exploratory analysis, scorecard development using logistic regression model and model monitoring features which are all used to provide the solution. Show less

• 

  ThirdEye Data

  Jun 2019 - Feb 2020

  Data Scientist

  Created an automated solution to detect anomalies in different sources of data related to server activities like CPU utilization from different machines and local servers as well as web server activity logs. Used threshold and forecasting approaches like robust z score and time series models to produce an upper and lower limits and anomalies are flagged based on the data points going outside the upper limits.Involved in adding features to some in house applications related to segmentation and predictive modeling. Show less

• 

  Protiviti India Member Firm

  Jun 2020 - Sept 2023

  Reviewed the Sanctions Screening Filter settings designed for customer and transactions screening for a middle east banking client by analyzing the filter performance on test data generated using Swift Sanctions Testing Services (STS) and running the test data past fircosoft filter settings (Firco Continuity for transactions, Firco Trust & Firco Screen and Keep for names). Swift STS is also used as a benchmarking tool to identify whether the filters are effective (Raising an alert against a sanctioned guy) and efficient (Missing a non sanctioned guy). Provided recommendations based on observed values on effectiveness and efficiency by tweeking rules and deactivating algorithms in fircosoft settings Show less Forecasted future sales of a major media enterprise by analyzing their past sales at a channel level and then redistributed the forecasted sales at a zonal level by analyzing the past distribution across 4 zones viz. East, West, North and South. Predicted the optimal coal blend that produces the coak with best quality with a low cost by analyzing the historical data of chemical properties of coal and coak and also the coal blends used for a major manufacturing client.Created a fully automated solution to produce a one month ahead forecast of freight rates of vehicles at a region level by analyzing the past trends of freight rates for a major manufacturing client.Delivered the phase 1 of an automated resume shortlisting use case for the HR department of a major petrochemical client in Qatar by extracting, cleaning and mapping raw texts from resumes against an exhaustive set of skill taxonomies followed by training a ML algorithm on the mapped dataset to predict the possible outcome of a CV i.e., whether it can be shortlisted for an interview or not. Show less

  • Deputy Manager - Data Science

    Jan 2023 - Sept 2023

  • Senior Consultant - Data Science

    Jan 2022 - Jan 2023

  • Consultant 3 - Data Science

    Jun 2020 - Jan 2022
• 

  Deloitte

  Sept 2023 - now

  Deputy Manager - Analytics & Cognitive (SAMA)

  Logistics Simulation Engine - Built simulation platform for logistics business users of a worlds leading electronics manufacturing company which enables changing existing transportation management business rules and run pass the historical order transactions through the rules & an optimization logic for career & mode selection to see the overall change in KPIs viz. Out Bound Cost, Cost Of Dissatisfaction & CO2 Emissions.Tools: Python (Pyspark, Airflow), AWS S3