Rupak Roy (Bob)

MicroMasters ~ Data Science

2021 - 2023

Massachusetts Institute of Technology, Cambridge

MIT Computer Science & Artificial Intelligence Laboratory (CSAIL) ~ Probability Redux, CLT, Law of Large Numbers, Hoeffding Inequality, Modes of Convergence, Parametric Statistical Models, Non- Parametric Models, Estimation, Confidence Intervals, and Delta Method, Confidence Intervals of the shift of shifted exponential random variables. Methods of Estimation. Linear classifiers, separability, perceptron algorithm, Maximum margin hyperplane, loss, regularization Stochastic gradient descent, over-fitting, generalization, Linear regression Recommender problems, collaborative filtering, Non-linear classification, kernels. Learning features, Neural networks, backpropagation. RNNs, Generalization, complexity, VC-dimension Unsupervised learning: clustering, Generative models, mixtures, Mixtures, EM algorithm Control Reinforcement learning, NLP and more.

Masters of Science ~ Information Technology

2010 - 2014

Data & File Structures, Object-Oriented Programming, Software Engineering, Operating System, Data Base Management System (DBMS), Analysis and Design of Algorithms, Data Communication, and Networking Object Oriented Analysis and Design(OOAD),Distributed Operating System, Embedded Systems,C# and .Net, Advanced Software Engineering

Masters ~ Project Management

2021 - 2023

PMP, Agile Scrum Master, PMO (Project Management Office) Lean Six Sigma Green Belt (LSSGB), Lean Magement PMI - RMP (Risk Management Professional) PMI - PgMP (Program Management Professional) Microsoft Project 2019

Masters ~ Commerce

2011 - 2013

International Business Environment, International Marketing Management International Business Finance, International Marketing Logistics, Trade Procedures Organizational Theory And Behaviour, Research Methodology and Statistical Analysis, Accounting for Managerial Decisions, Financial Management

Sr. Data Scientist, Advanced Analytics

2021 - Present

Amdocs, Data & Intelligence, Pune

• Assigned to/for data science oriented value proposition via marketing campaigns.
• Responsible for developing advanced ensemble, hybrid, genetic based auto-ml models.
• Leading a team of operations including data scientists to communicate and integrate the next gen. hybrid/genetic-based predictive models to the existing engine.

Data Scientist & Consultant

2018 - 2021

Data2Dimensions, Tampa, FL

• Designed the SDLC of machine learning, computer vision, and IoT projects.
• Outlining the scope of success with data science Strategy for the client in various industries of finance, production, inventory, marketing, law enforcement, etc.
• Delivered predictive data-oriented decision-making projects and other Ai solutions including corporate training with ML and Statistical Tools.

Data Science / Machine learning Trainer

2019 - present

Udemy, Ai Master-Class, Global

• Developed machine learning, big data and other statistical courses, also includes training and support to the candidates on the courses to tackle real life projects.

Sr. Engineer, Data Center

2014 - 2020

KommLabs, North-East, Shillong

• Handling 4 Projects combining 18servers from Nas to Central Database Server, Mediation Server as well as Store & Forward Server solely responsible for SDLC of the data mgnt. projects and analysis as well as troubleshooting and training client is crucial part of my job.
• Training all new colleagues in the insights and logistics of the organization including but not limited to the software, database, server applications and the analytics for the company procedures.

in brief

• Custom Ensemble modeling: Bagging, Boosting, AdaBoost, Voting, Stacking, Binding, Forecasting
• Ensemble Methods, Genetic Ai programming, Elastic Net, Radius Nearest-neighbor Regression, Auto-Feat Engineering and more.
• Hybrid Models, CNN-LSTM-MLP, Multivariate, Multi-Step Forecasting, Multivariate Parallel Input Series, Multi-Step Input Series
• Outlier Detection Models, Isolation Forest, MCD, LOF, OneClassSVM, DBSCAN.
• Synthetic Sample Generation: Adaptive Synthetic Sampling, SMOTE-ENN, SMOTE- TOMEK
• Auto and Semi-AutoML, Auto-keras, Optuna, Ploynomial Feat Engineering.
• AutoEncoders, T- SNE Anomaly/Fraud Detection, Fuzzy, BRICH, DBSCAN, Optics, Facebook, Neural Prophet, Thymeboost, Auto-TS-Enemble

and more.

DS Applications

• Developed Gui based Auto-Machine Learnig Data Analytics Kit v1_beta with 10+ power classifiers demo@ https://streamlit-roy.herokuapp.com mapped to knime/alteryx analytics.
• Developed End to end Anomaly Detection Systems with RCA Root Cause Analysis mapped to outlier.ai, anodot.
• End to end auto-optimized clustering using custom made autoelbow package @ https://pypi.org/project/autoelbow as well with Auto Time-Series Ensemble @ https://pypi.org/project/auto-ts-ensemble

and more.

DS Training

• Data Science, Ai, Big Data across platforms like Amdocs, Udemy, D2academics as well inother institutions with current ratings of 4.4/5
• Trained law & enforcement agencies, healthcare as well as corporate clients about the insights and logistics of data science & Ai.

Analytics

• Hypothesis Testing for accessing the project performance with various Probability distributions, Anova, Two Sample test
• Conditional probability re-balancing for Universal Control Group (UCG) to Universal Target Group (UTG)
• Applied pythagoras theorem to autotomate elbow for Clustering at production level to find the category of patterns for the product usuage.
• Eclat modeling to understand consumer purchase behavior.
• Perform Upper Confident Bound and Thompson Sampling, the best alternative to A/B testing for quick reinforcement learning.

and more.

Sales & Marketing

2021 - Present

• Profit dynamic pricing models via Competitors.
• Customer Retention Modeling.
• Identify At-Risk Customers to Improve Retention.
• Identify customers who where about to subscribe but left.
• Precise Targeting and Re-Targeting.
• Identify and forecast product usage.
• Sales & Demand Re-order Level Forecasting.
• Inception & Opportunity Forecasting via competitors.

Manufacturing & IOT

2018 - 2021

• Identify stages with similar performances and predict the quantification impact of cause and effect relationship to reduce the manufacturing failures as well system health.
• Outlining the scope of success with data science Strategy for the client in various industries of finance, production, inventory, marketing, law enforcement, etc.
• Automated data science process like anomaly detection, rca, forecasting, predictive and to the existing MLops and other integrated applications for real time connection between applications performance.

Anomaly Detection

2019 - present

• Outlier Profiling using Half Space Trees Method to to give a supervised perception from marketing to business operations to system health.

System Health

2021 - Present

at production level

• Automated the process of predicting and monitoring system resources with feedback loops for immediate re-training accross the big data jobs to have synchronized fail-safe resource allocation for marketing campaigns.

Identify and Automated
Personalized targeting to meet the individual needs.

2018 - 2022

at production level

• Applied state of the art hybrid deep learning forecasting models ensembled with traditional ml models to sucessfuly identify the patterns without compromising the deep packet inspection protocol.

Application Error Detection and Forecast

2020 - present

at production level

• Live Anomaly detection accross operations to give management fine decision making information for O&Ms as well with downtime forecasted dates with RCA.

Auto ML

2020 - present

at production level

• Developed and developed world’s one and only auto machine learning with orchestrated on-premises end to end pipeline that detects data pattern and triggers the respective ml learning stack of supervised, unsupervised, sequential and ensemble.

Customer Migration

2022 - present

at production level

• Sucessfully able to meet the business needs by reverse engineering churn concept with false negative metrics.