Summary
Overview
Work History
Education
Skills
Timeline
Generic
SANJIB BISWAS

SANJIB BISWAS

Bangalore

Summary

Result-driven professional; targeting challenging and rewarding opportunities in Data Science with a leading organization of repute

Overview

11
11
years of professional experience

Work History

Tech Lead (III)

Carelon Global Solution
Bangalore
03.2023 - Current
  • Title: Patient Readmission Prediction Model
  • Objective: Based on historical hospital admission data patient readmission prediction for Medicare, Medicaid & Commercial channel
  • Build a big data pipeline from SQL Server to AWS & Airflow Schedule
  • Ensure Model Scalability using Snowflake
  • Build Department wise prediction model (Logistic, SVM, Random Forest, XG-Boost, Neural Net) Score/Validation model result in production pipeline & deploy the same.

DS Product Architecture Lead

Larsen & Toubro Infotech
Bangalore
10.2020 - Current
  • Title: Autonomous forecasting (Johnson & Johnson)
  • Objective: Develop scalable univariate & multivariate forecasting modules which can be deployed & integrated with current software
  • Models & Statistical Test: Used: ARIMA, SARIMA, Auto Regressive with residual Error, GLM, Naïve Forecast, Theta Forecast, ARCH, LSTM
  • Regression With ARIMA Error (ARIMAX & SARIMAX), Ensemble Method (XG-Boost, Random Forest), Auto Regressive Poisson Model, Negative Binomial Model
  • Kwiatkowski–Phillips–Schmidt–Shin, Augmented Dickey Fuller, Ljung-Box Q Test, SHAPIRO Test, Rolling & Sliding Window validation
  • Title: Anomaly Detection (Johnson & Johnson)
  • Objective: Develop a scalable software solution which will identify & explain Anomaly in sales data at different TA Level
  • Models & Methods Used: Data Process & Model Development in Spark
  • Anomaly Identification using Isolation-forest, K-Means Cluster, CART, IQR Understand at different dimension level what is the type of anomaly (Historical, Data add/loss, Latest Data Deviation) Understand severity of anomaly & from where it originates (Customer, SD, Region, Territory) Understand Why anomaly is generating (wrong entry, customer missing) Title: Scenario Risk Assessment (Johnson & Johnson)
  • Objective: Identify risk factors of a trial/project in terms of time & cost
  • Assess the probability of success and failure in terms of simulated scenarios
  • Assess the likelihood of risk in terms of different what if scenarios
  • Models & Methods Used: Likelihood Estimation of Scenario
  • Quantify risk in from Bayesian stats
  • Monte Carlo simulation
  • Critical path Assessment (CPA)

Assistant Manager

WNS
Bangalore
- 05.2019
  • Title: Buyer Financial Risk Assessment Automation
  • Objective: QBE insurance provides insurance credit to their buyers who sell products to different retailers
  • If the retailers fail to return credit, QBE provides the insurance claim to the respective insured party
  • The objective of the project is to automate the entire evaluation process along with predicting which retailers are likely to fail in their credit payments that in turn might increase the claim amount for QBE
  • Models & Methods Used: Playing a key role in developing Web Scrap algorithm to build financial database from Yahoo finance
  • Build a comprehensive report on Turnover, Market Capitalization, Gross Profit, EDITA, PEG ratio etc
  • Automating the entire financial assessment report for scoring purposes
  • Financial risk Assessment based on credit history & current financial status
  • ML model to identify the probability of default for loan guarantee
  • Assessment of Loss ratio
  • Title: Identification of Catastrophe Claim
  • Objective: QBE insurance takes reinsurance to minimize risk from large catastrophic claims
  • The claim amount is divided between QBE and the reinsurance company based on the type of damage
  • Due to wrong coding of the damage category in some places QBE fails to claim back money from the reinsurance company
  • Objective of this NLP project is to identify which catastrophic claims are wrongly coded
  • Models & Methods Used: Computed vectored approach to find key words for Hurricane
  • Conducted TF-IDF (Term Frequency Inverse Documentary Frequency) analysis to identify key words for catastrophic events
  • Prepared SVM model based on count vectored feature development; validated models (Confusion Matrix, ROC) Treatment of imbalance target data set
  • Title: United Kingdom Claim Lifecycle
  • Objective: QBE handles lots of claim in a year and it takes huge time and manpower to evaluate a claim and come to a final decision making whether to accept or deny the same
  • The objective of the project was to build and deploy a predictive model that will predict the probability with which to accept or reject a claim with a click
  • Models & Methods Used: Rectified Multicollinearity and variable remove based on VIF
  • Built Gradient Boosting Models and led hyper parameter optimization
  • Development of Logistic Regression Model to forecast claim outcome
  • Optimum claim allocation based on financial viability & optimality
  • Assessment of outcome of claim settlement on the financial performance matrix
  • Creating Random Forest Regressor Model to determine claim allocation
  • Performing Categorical Variable Grouping based on Weight of Evidence & Information Value.

Senior Executive

NIELSEN
Bangalore
- 08.2016
  • Key Result Areas: Formulated market research report on retail product; performed data cleaning, transformation and created new variables
  • Led missing value prediction using different statistical approach; performed outlier detection & remedial treatment
  • Evaluated SKU wise sales, conducted root causes analysis on trend movement, market share assessment & sales estimation
  • Optimized number of detections considering client SLA, statistical methodological improvement for optimum results Performed ANOVA Testing along with conducting group analysis.

Sr. Project Officer

ICICI-Winfra (I-win advisory)
Kolkata
- 04.2015
  • Major Projects: Estimation of Share in Tourism Sector in Total GDP 4 for Northeastern States of India, Government of India Viability Assessment of Potato Flake Export Zone, WB, Government of West Bengal Key Result Areas: Prepared VAR Model & conducted analysis on the same
  • Identified Granger Causality to finalize causal relation
  • Led Input & Output Matrix Assessment Implemented estimation of Market volume (both National & International) Led market entry blocking strategy & identified ways to break it
  • Conducted casual effect analysis of some plant failure
  • Worked on Financial viability assessment of plant.

Analyst

Cognizant Technology Solutions
Hyderabad
01.2013 - 07.2013
  • Key Result Areas: Identified data consistency across large data sets; forecasted Time series
  • Performed data validation; conducted secondary research

Data Scientist

Commonwealth Bank of Australia
Bangalore
10.2022
  • Extract log report from autonomous tool.

Senior Data Scientist

Skoruz Technologies
Bangalore
05.2019
  • Title: Retail Loan Default Prediction (UK based Fintech)
  • Objective: Based On Retail Lending data build a scalable Expected Loss prediction model which can be deployed on cloud as software as services
  • EDA & Feature Creation based on the available variables (Annual Income, Address State, Avg Current Balance, DTI, Loan Grading & Sub Grading, FICO Score, Mortgage Term & Mortgage amount, Interest Rate) Weight of Evidence, Information Value, coarse classing Build PD Model (Logistic Regression, Random Forest)
  • Precession-Recall adjustment for imbalanced data
  • Validation using 5-Fold, F1-Score, ROC curve
  • Scorecard Building from PD model
  • Model Monitoring using PSI (Population Stability Index) LGD & EAD Model (Beta Distributed Regression Model Title: Available to promise & Inventory Optimization (Johnson & Johnson)
  • Objective: Objective of this project was predicting the shipment arrival time based on historical data along with that optimize the inventory in the supply chain network
  • Models & Methods Used: Generated lead time for delivery of medical devices line of pharma products Developed Random Forest, XGB model to predict the promised time
  • LPP is used for inventory Optimization based on Min Q, Min Max, Hybrid, Base & Periodic policy
  • Derivation of price elasticity model (Log-Log model).

Senior Consultant

IPSOS
Bangalore
  • Objective: Every organization spends huge amounts on marketing activity (TV, Radio, Facebook, Instagram, Newspaper etc.)
  • This project aimed at answering which marketing tactic derive how much sales by evaluating every marketing activity
  • The project also helped in calculating next year’s optimum financial allocation of budget on different marketing activities given the objective of the organization
  • Discussing and analyzing business questions with client/stakeholders
  • Directing the methodology & required variable transformation
  • Preparing market mix model, lead optimization & ROI calculation of media tactics
  • Working on Due-to & contribution assessment; providing suggestions on investment amounts in different marketing tactics.

Education

M.Sc. - Economics

University of Calcutta
01.2011

B.Sc. - Economics

University of Calcutta
01.2009

Skills

  • Deep Learning
  • Natural Language Processing (NLP)
  • AWS
  • ML Model Deployment & Dashboard
  • Technical Lead
  • Python & Pyspark
  • BFSI / Retail / Supply Chain/Pharma
  • Airflow
  • Big Data Pipeline
  • Data Science
  • Advanced Python
  • Advanced Excel
  • SQL
  • R
  • Problem Solver
  • Client & Stakeholder Management
  • Team Management
  • Soft Skills
  • Mentoring

Timeline

Tech Lead (III)

Carelon Global Solution
03.2023 - Current

Data Scientist

Commonwealth Bank of Australia
10.2022

DS Product Architecture Lead

Larsen & Toubro Infotech
10.2020 - Current

Senior Data Scientist

Skoruz Technologies
05.2019

Analyst

Cognizant Technology Solutions
01.2013 - 07.2013

Assistant Manager

WNS
- 05.2019

Senior Executive

NIELSEN
- 08.2016

Sr. Project Officer

ICICI-Winfra (I-win advisory)
- 04.2015

Senior Consultant

IPSOS

M.Sc. - Economics

University of Calcutta

B.Sc. - Economics

University of Calcutta

Similar Profiles

CREATE PROFILE
SANJIB BISWAS