ARUNABHA GUPTA

arunabhagupta11@[Link]
Contact No - 9903827152
Permanent Address: The Preserve, Flat 101, Belathur Colony, Bangalore-560067

My Belief: To learn continuously, contribute effectively and grow consistently both as an individual & as a team.
Past Experience:
TCS - 4.75 Years, Bangalore (Swiss Reinsurance (Life & Health), Deutsche Bank (Private& Commercial Clients (PCC) International))
IBM - 11 Months, Kolkata (DHL (Customer Analytics), XL (Claims Analytics))
Cognizant - 4.83 Years, Kolkata (Travelers (Claims Analytics), BNY Mellon, Nike (Order Fulfillment), Walmart, Belk)
PROFILE
 Design and build data pipelines for business use case which in turn populates a use case specific data lake
 Design and build data utilities for data cleaning, harmonizing and anonymization
 Design and build write back enabled applications with interactive widgets
 Carry out exploratory analysis on data science use cases
 Hands on experience on Sentiment Analysis, Text Mining, NLP, Unsupervised and Supervised techniques such as
regression, classification, clustering and other applied predictive techniques

TECHNICAL DETAILS
 Languages: Python, Pyspark, SQL, YAML, PostgresSQL , Spark ML
 Analytical Skills: Exploratory Data Analysis, Hypothesis Testing
Supervised and Un-Supervised Machine Learning Techniques such as Linear Regression, Logistic Regression,
Ensemble Techniques-Bagging, boosting, Random Forest, Natural Language processing-Sentiment Analysis, Topic
Mining, Neural Network, KNN,PCA
 Tools: Palantir Foundry (SLATE, CONTOUR, HUBBLE, CODE_REPO, CODE_WORKBOOK)
Cognos, Azure ML Databricks, Jupyter Notebook, PowerBI, GCP, AWS
PROFESSIONAL EXPERIENCE
Data Science:
Claims Customer International Address Parsing:
 Multilingual tokenization (Using Hidden Markov Model)
 Abbreviation expansion (Using Wordnet)
 Address language classification (Using the FTRL-Proximal method to induce sparsity)
 Numeric expression parsing
Health-Care Claims Customer insights:
 Analyzing customer feedback records for Topic Modelling. (Using Hierarchical Bayesian Model-Latent Dirichlet Allocation).
 Emotion Mining and getting Sentiment Scores.
Twitter Customer Sentiment Analysis:
 Sourcing data via spark streaming and twitter API, data cleaning
 Data understanding and pre-processing text by translating non-English comments using goggle translator, handling
emoji etc.
 Identifying different aspects in text data using analytical techniques such as topic modelling
Industry Categorization:
 Gathering data from client provided API via customized scrapper
 Cleaning data, preprocessing steps
 Data visualization and EDA to determine variable importance
 Feature engineering to create variables that were used in training model
 Model Selection (Used Random Forest and Stochastic Gradient Descent Model)
 Building machine learning data pipelines
Clustering - Statistically grouping furnaces with similar efficiency & finding factors affecting efficiency:
 Pre-processing data and cleaning of data
 Dimensionality Reduction using (PCA/LDA)
 Determining optimal number of clusters using Elbow curve
 Internal Cluster Validation using Silhouette score
Data Utilities:
Designing data utilities to ensure seamless ingestion of data from unstructured raw files.
 Data cleaning and data harmonization
 Data anonymization and combining multiple data files into a single source
 Snake casing using customized library
 Dataset homogeneity utility
 Dataset formatting utility
 Palantir Foundry platform specific tools used: CODE-Workbook for automation
 Build tool: gradle
 Versioning: GIT
Data Engineering:
Build Data pipelines on a scalable data platform.
 Sourcing data from unstructured/structured raw files
 Cleaning data as per industry standards
 Building data pipelines for the underlying business logic
 Optimizing the code for efficient memory utilization
 Explaining the technical implications in an understandable manner to end business users
 Designing Dashboards for business users to validate the changes
Data Pipeline Business Use Cases I have so far worked on
 Premium Validation and Anomaly detection
 Mortality Analysis
 Medical Reinsurance
 Smart Underwriting Risk Assessment
 Valuations
 Covid-19 Response
 Covid-19 What If analysis – In partnership with Palantir
 Palantir platform specific tools used: CONTOUR/HUBBLE/CODE-REPO
 Build Tool: Gradle
 Versioning: GIT
Dashboard Designing/Application Design:
Building applications on scalable analytics platform
Primarily responsible for developing interactive dashboards using Cognos/PowerBI for end business users.
Additionally I was able to scale up to developing front-end applications on SLATE - Palantir Foundry
Applications Contributed In:
 Actual vs. Expected & Data Freshness dashboard for SwissRe Life and Health
Dashboards Contributed In:
 Private and Commercial Clients International for Deutsche Bank Spain
 Claims dashboard for XL Catlin Group
 Order Fulfillment System dashboard for Nike US
 Insurance dashboard for Travelers Inc.
 Store Expansion Dashboard for Walmart US
 Tools: Cognos/PowerBI/Palantir Foundry SLATE
 Build Tool: Gradle
 Versioning: GIT

AWARDS AND RECOGNITION

ACADEMIC PROFILE
Degree University/Institution Year (CGPA)
Bachelor in Comp App WBUT 2010(83%)