Samiran Bera, PhD
Dr. Samiran Bera, PhD
| 📧 Email: samironbera@gmail.com | 📱 Mobile: (91) 75449 08874 |
| 🔗 Network: LinkedIn, GitHub, Medium | 🌏 Country: India |
📌 PROFILE SUMMARY
- PhD in Mathematical Modelling and Optimisation from IIT Dhanbad
- Senior Data Scientist with expertise in:
- Mathematical Modelling & Optimisation
- Descriptive, Diagnostic, Predictive, and Prescriptive Analytics
- Algorithm development – Classical, Tree-based & Heuristics
- Hands-on experience in:
- Fraud detection & Anomaly detection
- Customer segmentation & Recommendation systems
- Sentiment analysis, Cohort analysis, Market basket analysis
- Time series forecasting
- Artificial Neural Networks, Deep Learning
- Natural Language Processing
- Scheduling, Routing, Allocation, and Clustering
- Research & Publications:
- Published research papers in Elsevier, Sage & InderScience
- Presented POCs in international conferences (SOM & DSI)
- Authored book chapters in Springer & technical articles on Medium
- Developed a new Tree-based Branch and Cut algorithm
- Programming & Tools:
- Python, SQL, PowerBI
- Mentorship & Talent Acquisition:
- Mentored Data Scientists, MLOps teams, and interns
💼 WORK EXPERIENCE
| Position | Organization | Duration |
|---|---|---|
| Senior Data Scientist | Blackhawk Network, Bangalore | March 2022 – Present |
| Data Scientist | Blackhawk Network, Bangalore | March 2021 – Feb 2022 |
| Consultant (Data Science & ML) | GYPR Private Limited, Delhi | June 2020 – Jan 2021 |
| Senior Research Scholar | Indian Institute of Technology, Dhanbad | July 2016 – July 2019 |
| Junior Research Scholar | Indian Institute of Technology, Dhanbad | July 2014 – July 2016 |
🔍 MODEL DEVELOPMENT EXPERIENCE
| Category | Solution Developed |
|---|---|
| Data Analytics | Customer Lifetime Value (CLV), Cohort Analysis, Market Basket Analysis |
| Classification Models | Anomaly Detection, Fraud Detection, Churn Analysis |
| Clustering Models | Risk Profiling, Customer Segmentation, Recommendation Systems, Cellular Manufacturing |
| Regression Models | Demand Forecasting, Price Forecasting, Drift Detection |
| Language Models | Sentiment Analysis, Topic Modeling, Feedback Analysis |
| Optimization Methods | Classical Methods, Tree-Based Search, Heuristics & Meta-Heuristics |
🛠️ TECHNICAL EXPERIENCE
| Field | Details of Expertise |
|---|---|
| Programming | Python, R, SQL, MATLAB, GAMS, CPLEX |
| Tools | PowerBI, Anaconda, Jupyter Notebook, PyCharm, VSCode, Git |
| Analysis & Learning | Data Exploration, Quality Checks, Analytics, Diagnostics, Profiling & Visualization |
| Machine Learning | ML & Deep Learning, ANN, NLP |
| Statistical Tests | A/B Testing, Champion Challenger Test, KS Test, Drift Detection Test, Back Testing |
| Algorithms | Linear & Logistic Regression, Decision Trees, Gradient Boost, XGBoost, Random Forest, Isolation Forest |
| Neural Networks & Time Series | CNN, RNN, LSTM, ARIMA, DART, Prophet |
🎓 ACADEMIC QUALIFICATIONS
- PhD in Mathematical Modelling & Optimization, Indian Institute of Technology Dhanbad (2023)
- M.Tech in Industrial Engineering & Management, West Bengal University of Technology (2014)
- B.Tech in Computer Science & Engineering, West Bengal University of Technology (2012)
📄 RESEARCH PUBLICATIONS
- Bera, S., Das, S. K., & Ghosh, J. (2021). An efficient routing in wireless sensor network: an application of grey wolf optimization. Nature-inspired computing for smart application design, 143-168.
- Bera, S. (2020). An application of operational analytics: for predicting sales revenue of restaurant. In Machine learning algorithms for industrial applications (pp. 209-235). Cham: Springer International Publishing.
- Bera, S., Das, S. K., & Karati, A. (2020). Intelligent routing in wireless sensor network based on african buffalo optimization. Nature inspired computing for wireless sensor networks, 119-142.
- Bera, S., Das, S. K., & Karati, A. (2020). Intelligent Routing in Wireless Sensor. Nature Inspired Computing for Wireless Sensor Networks, 119.
- Kuthambalayan, T. S., & Bera, S. (2020). Managing product variety with mixed make-to-stock/make-to-order production strategy and guaranteed delivery time under stochastic demand. Computers & Industrial Engineering, 147, 106603.
- Kuthambalayan, T. S., & Bera, S. (2020). A review of the literature on mixed make-to-stock/make-to-order production systems: major findings and directions for future research. International Journal of Services and Operations Management, 37(3), 372-406.
- Bera, S., Chattopadhyay, M., & Dan, P. K. (2018). A two-stage novel approach using centre ordering of vectors on agglomerative hierarchical clustering for manufacturing cell formation. Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture, 232(14), 2651-2662.
📝 TECHNICAL ARTICLES ON MEDIUM
- Backtesting Strategy for ML Models – published with Towards AI
- Feature Selection — Using Genetic Algorithm – published with Analytics Vidhya
- Mutation — Break Free from Stagnation – published with Analytics Vidhya
- Transportation Problem — Solve using Genetic Algorithm – published with The Start Up
- Customer Segmentation using K-Means Clustering
- Formulate Vehicle Routing Problem— Using GAMS
- Crossover Operator — The Heart of Genetic Algorithm
🏆 AWARDS & ACCOMPLISHMENTS
🏅 Silver Medalist, M.Tech, West Bengal University of Technology (2014)
🎓 AICTE Government Scholarship, M.Tech (2012 – 2014)
🎓 MHRD Government Scholarship, PhD (2014 – 2019)
🏆 Grand Prize Winner, Hawkathon 2022 (Best DS/ML Application POC)
🏅 Best Paper Award, FOSET National Conference (2012)
📊 PROJECT EXPERIENCE
Fraud Prevention & Risk Analytics
- Developed Direct-to-Consumer Fraud Prevention model: Designed a fraud detection model for direct-to-consumer sales using Python, Scikit-learn and XGBoost. Improved chargeback and approval rates by optimizing precision-recall trade-offs.
- Developed brand-specific Fraud Prevention model: Built fraud models incorporating brand-specific risk signals using supervised learning techniques. Used Scikit-learn and feature engineering to enhance fraud detection accuracy.
- Improved existing Fraud Prevention model by developing new Predictors: Enhanced fraud models by introducing new predictive features through SHAP, LIME and Information Value based feature selection and optimized feature combination. Improved approval rates and lower chargeback resulted in better fraud classification.
- Developed infrastructure for champion-challenger test, performance drift analysis, and backtesting: Designed a framework for continuous model validation, incorporating statistical hypothesis testing (Kolmogorov-Smirnov, Chi-square tests) to detect performance drift and optimize fraud detection.
- Developed Anomaly Detection model using Isolation Forest: Implemented anomaly detection using Isolation Forest and Local Outlier Factor to identify suspected activity. Tuned contamination parameters to reduce false positives and detect fraudulent transactions more effectively.
Customer Analytics & Recommendation Systems
- Developed POC for Customer Segmentation & Recommender System using RFM Analysis: Built a customer segmentation model using RFM analysis with K-Means clustering. Implemented collaborative filtering to improve targeted marketing and customer engagement.
- Developed POC for interactive voice response for customer feedback: Created an IVR-based feedback system and applied NLP sentiment analysis to extract actionable insights on order decision.
- Interpret Comments and Annotations using Natural Language Processing: Developed an NLP-based system to analyze reviewer comments on order approvals/declines, extracting key decision-making factors. Used Named Entity Recognition (NER), sentiment analysis, and topic modeling with Python (NLTK, spaCy) to rectify order decisions.
Data Visualization & Forecasting
-
Collaborated with Risk Analytics Team to develop PowerBI dashboard: Built an interactive PowerBI dashboard integrating SQL data pipelines. Optimized data visualization for fraud monitoring, improving decision-making efficiency. Helped analytics team to develop the dashboard.
-
Developed time series forecasting model of gift cards sales volume for Holiday season: Implemented a time series forecasting model using Facebook Prophet, ARIMA, and XGBoost-based regression. Achieved significant improvement in MAPE, enhancing holiday sales predictions.
Optimization & Mathematical Modeling
- Developed mathematical model to determine production strategy under demand uncertainty: Designed a stochastic optimization model using Mixed Integer Non Linear Programming to optimize production decisions under fluctuating demand. Improved cost efficiency and resource utilization.
- Lead Time Optimization using Mathematical Model: Developed a deterministic mathematical model using Integer Programming to minimize lead times. Achieved significant reductions in delivery delays.
Algorithms, AI-based Optimization & (Meta) Heuristics
- Integration of Greedy Algorithm with Branch and Cut Algorithm for faster convergence to optimal solution: Combined Greedy Algorithm with Branch and Cut to accelerate convergence in combinatorial optimization problems. Improved computational efficiency in solving large scale MINLP problem.
- Hybrid Hierarchical Clustering Algorithm to optimize parts movement to machine: Developed a hybrid clustering approach combining Hierarchical Clustering with COVR to optimize parts allocation, reducing intra-facility transportation costs and improving workflow efficiency.
- Transportation Problem using Genetic Algorithm: Applied Genetic Algorithm to optimize transportation costs and allocation efficiency in supply chain networks. Used fitness function tuning and adaptive mutation for improved convergence.
Interests & Hobbies
- Travel Photography
- Technical Writing
- Epoxy Resin DIY Projects
Languages
- English: Fluent
- Hindi: Fluent
- Bengali: Fluent (Native)
Connect with me
| Field | Details |
|---|---|
| Name | Dr. Samiran Bera, PhD |
| samironbera@gmail.com | |
| Website | https://samiranberahaldia.github.io/portfolio/ |
| Social | LinkTree |
| 🚀 **View My GitHub Projects | Contact Me** |