Ritesh Kumar

Data Science Professional

Data Science Leader with over 15 years of experience delivering multimillion-dollar value for Fortune 500 companies. Expert in machine learning and AI, transforming complex data into actionable insights. Solves business challenges using advanced technologies. Leads high-impact client projects, ensures data governance and compliance, optimizes workflows, and drives cross-functional collaboration to achieve strategic business objectives.

Projects

  • UNGA Topic Analysis: Analyzed speeches from the UN General Assembly using LDA and Topic GPT, revealing the distinct strengths of each model. LDA identified overarching themes such as peace and security, providing a wide-angle view of global concerns. In contrast, Topic GPT delved into more nuanced issues like climate change and human rights, offering a richer, more detailed perspective on international discourse. Read More

  • Algorithmic Trading with Deep Learning for BankNifty: Developed an advanced algorithmic trading system leveraging CNN to forecast minute-by-minute price movements of the BankNifty index. The system achieved notable predictive accuracy, and showed promise for profitability in trading under live market conditions. Read More

  • Salesperson Segmentation for Revenue Optimization: Implemented k-means clustering to segment salespersons based on performance metrics. The analysis revealed two distinct clusters: one group making fewer but high-value sales, and another making twice as many sales but with significantly lower average transaction values. This insight led to tailored training and resource allocation strategies, resulting in a 15% increase in overall sales revenue.

  • Netflix Performance Metrics Dashboard: Conducted in-depth data analysis and built a dynamic Power BI dashboard, leading to a 25% boost in Customer Satisfaction (CSAT) during weekends.

  • LLMs in Power BI: Engineered LLM-powered text summarization and topic-extraction in Power BI, integrating Gemini and GPT for dynamic, filter-responsive summaries in an Agent Feedback dashboard, overcoming technical limitations through innovative text-to-image conversion.

  • Media Mix Models for Maruti Car Sales: Led a cross-functional team to develop media mix models for eleven Maruti car models, analyzing over 50 sales-influencing factors, including seasonality, the impact of COVID-19, and marketing expenditures. Rigorous exploratory data analysis revealed key trends, leading to the adoption of Linear Regression combined with Box-Cox transformation for 8 models, delivering optimal predictive accuracy. The solution, securely deployed on AWS, empowered Maruti’s marketing team to forecast sales outcomes, with the project progressing smoothly through regular client reviews and feedback.