AYUSH KUMAR

ACADEMIC QUALIFICATIONS
Year Degree/Board University/Institution %/CGPA
2023 Post Graduate Diploma in Business Analytics IIM Calcutta, ISI Kolkata, IIT Kharagpur -
2019 B. Tech Mechanical Engineering IIT BHU (Varanasi) 6.8/10
2014 Class XII, CBSE Delhi Public School, Patna 91.8%
2012 Class X, CBSE Delhi Public School, Patna 10/10
AWARDS & ACHIEVEMENTS
Institute Rank  Currently ranked in Top 10 percentile in PGDBA 23 in academic performance among 60 students
CFA-Level 1  Cleared CFA Level 1 with 90+ percentile; Scored in highest percentile bracket in 9 out of the 10 subjects
 Ranked 1st among 2.5k+ participants in IIM-B Vista 2022: Data & Beyond case competition, won INR 50k
Data Science  Ranked 1st among 1.5k+ participants in IIM-B analytics competition on customer acquisition, won 12k
Competitions  Ranked 3rd in analytics competition at National Student’s Space Challenge organized by IIT KGP
 Ranked 5th among 2k+ participants in Data Science Hackathon organized by IIT Kharagpur and Redbus
 Ranked 1st in Marketing competition organized by IIM-C Alumni Chapter, won bitcoin worth 25k
Business
 Ranked 3rd among 1k+ participants in competition on pricing strategy by IIM-B & TheMathCompany, won 30k
Competitions  Ranked 3rd in Consulting case competition on Healthcare Information System organized by IIM Rohtak
WORK EXPERIENCE (29 months)
Associate Data Analyst United Health Group - Optum Noida (May’19 – Oct’21)
 Analyzed risks for 10 million customers in Fortune 500 firms by assessing the profit-making programs
 Captured suspects who could be potential risk bearers for our insurance products by developing 5 models
Risk Analysis  Delivered 5 dashboards per week and assisted financial reporting of claims compensation via Tableau
 Generated additional $15Mn+ revenue for our clients by identifying new diseases based on existing condition
 Fulfilled 40+ weekly.deliveries to the US and India clients using automated scripts with 90% + accuracy
 Devised a 300 TB Big Data pipeline valued at $15 Mn in downstream processes and presented to VP, UHG
 Spearheaded migration of codes and processes to Big Data env from SQL Server ensuring no redundancy
Big Data Analytics
 Processed and Analyzed 200TB data using Hive; automated Sqoop process to migrate this data to SQL
 Ascertained end-to-end product ownership of Health programs developing PySpark ETL pipelines
 Increased code productivity by 85%, decreased defects by 90% by reengineering 50k lines of codes
 Enhanced the Big Data memory occupancy by 39% by efficiently migrating the processes and the tables
Process Optimization
 Overhauled and Optimized 10+ production processes with 20+ modules by designing 4 dependent env
 Conserved 10+ resource hours/week by automating 15 big data processes using centralized oozies
 Negotiated with 5 cross-functional teams based out of US and India to ascertain change impacts in dev
Stakeholder
 Followed agile methodology and maintained transparency to align 40 stakeholders during dev tasks
Management and
 Collaborated with project leads in migration phase for need assessment, design precision and timeline est.
Talent Development
 Conducted 10+ technical knowledge transfer onboard sessions for 20 hires in Buddy-Connect program
 Awarded 9 Bravos (UHG’s Recognition System) worth Rs 1 lac for performance by clients & colleagues
Achievements  Awarded by external delivery managers for 95%+ accuracy in resolving code defects for 2 consec. years
 Appreciated and recognized by delivery managers and Fortune 500 clients for timely deliverance of work
ACADEMIC PROJECTS
 Built a deep learning model to classify multi-class surface defects (cracks, scratch, holes etc.) on steel sheets
Steel Surface Defect
 Used Transfer learning on VCG16, .ResNet101, DenseNet121 model to train on 7k+.steel surface images
Image classification
 Achieved maximum accuracy of 90.3% and F1 score of 80.6% using Xception architecture as the model
 Developed a model to classify the news article into real or fake based on content of the news and metadata
Fake News
 Processed data using stemming; vectorized using TF-IDF; used Logistic Regression, RF, AdaBoost and XGBoost
Classification
 Performed hyperparameter tuning using randomized grid search to achieve best F1 Score of .82 for XGB
 Predicted S&P 500 stock returns based on past 8 years of data using time series forecasting techniques
S&P 500 Return
 Checked stationarity using ADF test, analyzed ACF-PACF plots for ARIMA order, Ljung -Box for residuals
Forecasting
 Employed ARCH and GARCH models to forecast volatility of returns & achieved a lowest AIC value of -2872
End-to-End  Designed database for an Auto-Insurance firm using E-R model, defined attributes, entities & relationships
Database Design  Identified functional dependencies, Normalized tables to BCNF; used SQL and dummy data to create tables
ADDITIONAL PROJECTS
 Built a model for predicting likes of posts to mine insights for digital marketing, used 30k+ posts and metadata
Instagram Likes
 Compared Linear Regression, XGBoost on user-meta data features to achieve RMSE of 1000+ using XGBoost
Prediction
 Extracted image features using CNN & text features using CountVectorizer to further reduce RMSE to 350+
EXTRA CURRICULAR ACTIVITIES
 Collected, maintained and managed funds worth Rs. 1.7 lacs+ for student driven clubs and activities
Treasurer, PGDBA
 Handled fund disbursement worth Rs 90k+ with IIMC Treasurer, during the Annual Conclave, Trilytics
Sponsorship Team  Laid the foundation of institutes' 1st ever analytics competition hosting 350+ colleges with 1.7lac+ outreach
Conclave, Trilytics  Bagged sponsorships and negotiated PPIs for winners, in events having footfall of 2k+ participants pan India
Theatre Club,IIT BHU  Won 1st prize at cultural fest of IIT BHU; Represented college in Mood Indigo and Inter IIT in street play
TOOLS & LANGUAGES: Python, Hive, SQL ELECTIVES: Supply Chain Analytics, Pricing & Revenue Opt., Games & Information