Data Scientist
Resume:
RISHABH SONI
Data Scientist/Analytics Manager
C-8/61 A,
Keshav Puram, Delhi
***********.***@*****.***
PLATFORMS/TOOLS
R, Python
SAS/SQL
Hive/Hadoop, Pyspark
Excel/Power Point
Alteryx
Basic VBA, Tableau
TECHNIQUES
Regression(Linear/Logit/Gamma)
Regularization(Ridge/Lasso)
Decision Tree
PCA/Factor analysis/T-SNE
SGD Classifiers
Random Forest, Extra Trees
Boosting (GBM, XGB, Light-GBM)
K-Means/Modes/Hier. Clustering
Market Basket Analysis
Markov Chain Monte Carlo
Stacking/Blending Ensembles
K-Nearest Neighbors
Support Vector Machines
Feed Forward Deep Learning
Signal Processing(FFT/Wavelet)
Non Linear Time Series, ETS
Text Mining, Word2vec
GLRM, Truncated SVD
Dynamic Programming
Survival Analysis
Kalman Filter
Hypothesis Testing
Meta Bagged Models
EDUCATION
August 2006 – June 2010
B.Tech in Electrical Engineering from IIT Roorkee
CGPA: 8.72/10
CAREER ACHIEVEMENTS
Analytics Manager, UHG-Optum
December 2015 – Present
R&D initiatives
Developed signal processing capability on EEG data to predict seizure. Built a signal processing code library which can be utilized for any signal data. (Paper presented at international healthcare conference - IIM Ahmedabad)
Developed claim fraud rules by association mining and member clustering to identify wrongful diagnoses, procedure bundling fraud etc.
Built text mining capability which was utilized to enhance search capability of internal website.
Detect claims with high risk of appeals in M&R space. This is being replicated for other areas.
Project Delivery
Build network optimization algorithm to determine next best provider to meet CMS requirement of having atleast 90% member coverage for all specialties.
Detect providers with high chances of having an incorrect record. Model outcome utilized for campaigns (>60% hit rate for 2 campaigns done).
Detect ER visit for asthma patients using internal data supplemented by external data such as air pollutant, weather changes etc. (Paper presented in internal UHG conference)
Senior Data Scientist, Novartis
June 2015 – Nov 2015
Built a model for patient allocation in trials based on various constraints such as cost and time. Created a risk score to assess the performance of investigators in a clinical trial.
Senior Analyst, Boston Consulting Group
August 2014 – May 2015
Assess the effectiveness of promotions run by an Indian retailer and built a promo code library which can be utilized in future.
Analyzed transaction data for a Thai Retail Chain, identified B2B and B2C transactions, and created product association rules by region which were delivered in form of an excel tool.
Performed customer segmentation for an Indonesian Telecom by ARPU to understand customers’ needs/demands. Market basket and correlation analysis was used to bundle various packets for cross-sell.
Senior Analyst, Inductis - EXL
November 2010 – July 2014
Led the initiative to bring credit card submissions (Big Data, >20 TB) into HDFS and warehouse. Performed rigorous UAT to ensure data sanctity.
Conceptualized and created various Data Quality Reports to understand data utility. This assessment led to projects such as small merchant risk scanner& location based targeting.
Developed and implemented the Merchant Exposure Calculation (PIT) logic. Created dashboards to estimate reserves based on the merchant exposure
Project Manager overseeing the development of Exposure Management Capability. Interacted with global business partners to build wireframe for the tool.
Worked directly with senior leadership to size two major initiatives: Transaction amount/Portfolio based Pricing and small merchant portfolio revamp using external data.
Build strategy to revamp reporting and analytics of Fortune 300 Insurance Company with focus on organization restructuring and data management. Interviewed many leaders and SMEs to understand their requirements and current gaps in their analytics process.
Enhance marketing strategy of a top credit card issuer by analyzing customers’ affinity towards email and online channel.
Developed and Implemented offer type level customer segmentation using predictive modeling. Assessed the performance of various big campaigns using look-alike methodology
Built new suppression logic for targeting existing sellers which led to 100K additional leads. Created logic for trigger based targeting such as revenue spikes, increase in stores etc.
MISCELLANEOUS
Led a project sharing initiative for the consulting team at Inductis.
Member of recruitment team at Inductis/UHG.
Conducted R, SAS, Hadoop/HiveQL& ML training sessions.
LANGUAGES: Hindi (native), English, Punjabi (basic)
INTERESTS: Reading Novels, Travelling, Amateur pool and Tennis player
Contact this candidate