SUMMARY:

• Nineteen years of experience in Architecting, developing and Managing teams in Enterprise software solutions including Machine Learning, Artificial Intelligence (AI), Big Data technologies, Hadoop, Machine Learning Supervised and Unsupervised Learning, Text Analytics and Unstructured data analysis. Worked on various Recommendation Engines and several Java/J2EE, .Net, TIBCO based platforms.
• Work experience includes assignments as Senior Big Data Executive, Data Scientist/Architect, Enterprise Architect, Mentor at Bank of America Merrill Lynch, Johnson & Johnson, HSBC, Mayo Clinic, American Express, DuPont, Goldman Sachs, Federal Reserve Bank (NY), Merrill Lynch, Bank of America, Bell South, Accenture

TECHNICAL SKILLS:

Languages: Python, R, Scala, Pig, Hive, Java/J2EE, JSP, C#, ASP.Net

Database: HBase, M7, HDFS, Oracle, DB2, Sybase, SQL Server, JDBC, ODBC, Toplink, Hibernate, SAP R/3

Development: Hadoop, MapReduce, Spark, Mahout, Data Modeling, Yarn, Zookeeper, J2EE, EJB, SOAP, REST

Tools: Used: Confidential, Eureqa, H2O.ai, Scikit - learn, Pandas, Numpy, Neo4J, Graph Database, Tableau, MemCache, RCurl, RDD, JDK, Tibco, Matlab, Revolution R, Flume, Sqoop, Unica, Oozie, Datameer, Platfora, MapR, Cloudera, BigInsights, Azure, AWS, EC2, EMR, Pandas, Numpy, Anaconda, Spyder, Jupyter, Mapplotlib, Seaborn, Lucene, Elastic Search, Solr, Sklearn

Data Science: Supervised Learning, Unsupervised Learning, Deep Learning, Reinforced Learning, Regression, Lasso, Ridge, KNN, KMeans, Clustering, Neural Networks, SVM, Na ve Bayes, Fuzzy Logic, NLP, NLTK, Spacy, Topic Modeling, Decision Trees, Random Forest, Bagging, Boosting, GBM, AdaBoost, TensorFlow, SkFlow, Algorithmia, CNN, Deep Learning, BigInsights, ARIMA

EXPERIENCE:

Confidential, New York, NY

Customer Facing Data Scientist

Responsibilities:

• Ensured alignment with key technology and business stakeholders across globally diverse, Agile teams.
• Helped Bank of America through building predictive models in FX, Fixed Income, Investment Banking, Research, Commercial Lending, Wholesale Credit, Client Relations, M&A.
• FX Volume Prediction: built time series models to predict daily FX volume for CLS at a minute level using data from primary exchanges - EBS, Reuters, BofA volume, bid and ask rates, spreads, VWAPs, simple and exponential moving averages, order book entries, etc. Used Bollinger Bands, MACD, market events, holidays for EUR/USD, USD/CAD, etc currency pairs. MASE values were impressive compared to a naïve model.
• Customer Attrition Prediction: built a highly successful customer attrition predictive model with 80% accuracy on FICC electronic trading from Bloomberg terminals using time series, feature engineering with financial ratios, etc.
• Capital Review Committee Revenue Prediction: predicted yearly revenues for years 1 to 3 for the bank on 16 products ranging from Treasury, Advisory, Credit to FX, and had beat bankers estimates.
• Funded Loan Growth Prediction: developed predictive models for funded loan growth for the Corporate Banking group at Bank of America and improved prediction probability six times. Found key drivers and early indicators.
• Worked with Balyasny Asset Management (BAM) hedge fund, JPMC, TD Bank, GRA (Global Risk) at BAML.
• Built custom machine learning models on large datasets in use cases such as optimal capital allocation, commercial loan growth, customer attrition, market trend prediction for the bank.
• Built multi-class sentiment analysis models on bank’s research reports using NLP and Spacy.
• Moved Machine Learning projects into production and created tangible value for the firms.
• Brought business insights showing feature interactions in ratings tables, prediction explanations.
• Built several workflows that combined data preprocessing steps with feature engineering, feature selection, model selection, hyper-parameter tuning, model stacking, blending, using cross validation to avoid overfitting, validating models with lift charts and ROC curves, explaining insights through feature importance analysis, partial dependency plots. Handled class imbalance and large datasets. Explored human - machine hybrid approaches.
• Captured trends, seasonality patterns through time series models such as ARIMA, used lag variables and sliding window techniques, feature engineered variables through iterations.
• Analyzed unstructured text in analyst reports, built sentiment analysis using TFIDF, NLP, Spacy.
• Balanced algorithm accuracy over speed in XGBoost, Random Forest, GLM, ENet Blender, Logistic.
• Worked with bank regulators on variable stress testing, parameter sensitivity analysis.
• Built challenger models for BAML regulators, the Model Review Management group, a three month long process, with variable stress testing, hyper-parameter sensitivity analysis, out of time validation, and model deployment.
• Helped Humana insurance with Marketing mix optimization, Emergency Room attendance estimates.
• Developed Oil recovery models for Hunt Oil, and transport ETA predictions for Rail Inc. and BASF.
• Handled large scale transactional, trading, loan, hospital, transportation, oil production data.
• Evangelized Artificial Intelligence, Machine Learning through presentations, online webinars, blogs.
• Wrote popular blogs on Machine Learning and received company’s special award on content creation.

Environment: Python, R, SKLearn, Time Series, ARIMA, Multiclass, Anomaly Detection, Feature Engineering, Imbalanced data, SQL, Hive, Hadoop, Tableau, Spacy, NLP, Spark, Confidential, Eureqa, Nutonian

Confidential, Princeton, NJ

Data Science & Big Data

Responsibilities:

• Developed real-time, transaction screening SaaS software that reduce false positives and false negatives and detects risk before a financial crime happens. Intelligent, data science driven SaaS software on the cloud or on-prem for transaction/customer screening, insider threat, FCPA.
• Driven by machine learning, natural language processing, text analytics, link analysis using graph databases.
• Linked hidden PII data such as email ids, phone numbers, addresses, friends, relatives, work relations, etc.
• Used public source data, sanctions list, OFAC lists, entity resolved private vendor, data from KYC processes, from correspondent banks. Used 3 Vs: volume, value, velocity of transactions on SWIFT, Chip type transactions.
• Did holistic risk assessment and risk intelligence. Built prescriptive systems where systems aid decision making.
• Highly flexible and configurable platform with easy integration capabilities into Hadoop based data lakes as well as advanced engines based on data science and AI.
• Developed predictive models using cluster analysis, anomaly detection, and feature selection through PCA for dimensionality reduction using their large transactional and KYC datasets.
• Helped bring external data from vendors such as Factset, NexisLexis.

Confidential, Los Angeles, CA

Responsibilities:

• Helped this retail-chain stay competitive through cutting edge pricing analytics models from large scale data.
• Developed demand prediction models using supervised machine learning models such Random Forest Regressor, Extra Trees Regressor, SVM, XGBoost, GBM, Logistic Regression.
• Worked on large sets of sales, transaction, marketing, and pricing data. Conducted data analysis/cleaning, predictive analytics, handled inventory sold outs, generated price elasticity curves.
• Applied price optimization using advanced price optimization libraries in Python and dynamic pricing models for revenue management