Prafulla Shukla

Data Science Leadership:

• Led the delivery of best-in-class AI-ML based demand forecasts for large scale retailers (Fashion, Apparel, CPG) for merchandise optimization and pricing/promotion strategies.

Team Leadership and Organization Building

• Led a team of 3-6 through project delivery while grooming them for upskilling, adopting best practices, task management, prioritization and overall professional growth.
• Hiring the best talent through rigorous assessment and on-boarding with step-up training

Project and Stakeholder Management:

• Scoped, planned and delivered data science projects by aligning requirements and results with top business stakeholders.
• Continuous assessment and communication of model performance, challenges and next steps

Data Science RnD

• Continuous enhancements and additions in the existing AI/ML pipeline in CI/CD fashion
• Integrated new algorithms such as Bayesian ML hyperparameter tuning, bias correction, new item forecasting etc.

Digital Transformation

• Established 360° analytical performance reporting
• Developed self-serve Tableau dashboards for all business functions and their KPIs such as funnel conversion, AOV, active customers, GMV, margin, delivery time, replenishment etc.
• Delivered organization-wide analytics training, driving efficiency and data-driven approach

Data Science Solutions

• Identified customer profiles by synthesizing 360° (online & offline) behavioural aspects to improve campaign design and CLTV management
• Empowered B2B e-commerce platform with product discovery/recommendation widgets
• Built end-to-end automated, in-house CRM system with personalized recommendations.
• Deployed acquisition and churn prevention models, demand forecasts and delivery optimization models

Collaboration, Stakeholder Management and Data Leadership

• Developed long term data strategy with top leadership for Ecom business in collaboration with marketing, merchandising, membership and store operations teams
• Enabled marketing, merchandising and membership teams with analytical reporting and predictive models to achieve higher customers conversion, faster sell through and improved acquisitions
• Equipped team members with automation and analytical tools and techniques to become more agile and robust in their day to day work.

Key Client Work

• Developed an ML based Predictive Maintenance model (Industry 4.0) for largest automotive parts manufacturer, to prevent downtime by predicting key component failure in advance for a rare but high impact failure event
• Built a Gameplay Viewership prediction and Optimization solution (mix of Poisson Regression and LPP) for a leading media company in India, improving of ~20% in TV ratings
• Built a Churn Prevention and Attribution Strategy for a AVOD platform of a media house, combining ML and Statistics giving customer level drivers of churn
• Delivered long term, world-wide, monthly Natural Gas Price predictions for largest Asian Supplier by building Demand and supply forecasts and custom optimizing algorithm supply gap
• Built a scenario simulation and execution tool in Excel VBA to foresee and plan for all possibilities in demand/supply

Capability and Organisation Building

• Capability Development/Analytics Training for the internal analytics team (mix of fresher, middle and upper management)
• Delivered extensive analytics training and helped client in building a strong analytics and data science team

Business and Org Development

• Built cross industry use-case POCs
• Lead the talent development via building and delivering analytics trainings within team

Data Science Modules

• Property and Casualty Insurance- Share of Wallet (premium) Estimation
• Developed an end to end data quality management and model selection tool in R
• Created white papers and sales material on these applications of predictive analytics

• Risk and Fraud prediction modeling for Online Pay-day lending business for sub-prime market
• Built ensemble models to estimate income from lending; including models for PD, pre-payment, conversion and fraud propensity
• Predicted the Customer in Financial Distress (CFD) i.e Insolvency etc., overlaying atop fraud models Identified a highly risky segment with ~50% fraud rates vs. ~2% overall

• Data Analysis and Automated Strategic Segmentation of the customer base using BASE SAS, using data steps, procedures and macros; increased efficiency from a day to ~30 minutes