Projects

1. Relevance-to-Revenue Engine (Pricing & Explainability)

A decision intelligence system that combines demand modeling, revenue optimization, and learning-to-rank with GenAI explanations, translating complex pricing and relevance decisions into clear, human-understandable insights. Read More →

Decision Intelligence Pricing Analytics ML NLP Explainable AI GenAI Mistral-7B Streamlit Production-Ready

Key Impact:

Estimated price elasticity quantified for listings
Expected revenue uplift trade-offs surfaced in narratives
Booking probability insights contextualized for business decisions

2. Intelligent Document Experience Assistant (IDEA)

A document intelligence system that converts unstructured enterprise documents into structured summaries, key themes, and explicit risk signals, enabling faster comprehension, consistent review, and decision-ready insights through responsible use of LLMs. Read More →

Document Intelligence NLP Applied ML LLMs Prompt Engineering Risk & Compliance Explainable AI Mistral-7B Streamlit

Key Impact:

Reduced manual document review effort by surfacing concise summaries
Explicit risk and gap signals extracted to support compliance
Improved consistency of interpretation across stakeholders

3. Product Adoption & Churn Intelligence Assistant

An agentic decision system that analyzes customer usage signals to surface feature adoption opportunities, actionable enablement playbooks, and explainable churn risk, translating raw telemetry into CSM-ready guidance. Read More →

Product Analytics Agentic AI Explainable Systems Customer Intelligence Python Streamlit Modular Architecture

Key Impact:

Under-utilized feature opportunities surfaced with clear business rationale
Churn risk signals consolidated into explainable Low / Medium / High classifications
Reduced manual analysis for Customer Success and Product teams

4. Marketing Intelligence Engine (Time-Aware Propensity Modeling)

An end-to-end, leakage-free machine learning system for predicting term-deposit subscription under real-world campaign dynamics—combining rigorous temporal evaluation, imbalance-aware modeling, and operational metrics to support realistic marketing decisions. Read More →

Applied Machine Learning Time-Series Evaluation Campaign Analytics Classification XGBoost Model Validation Imbalanced Learning Explainable ML

Key Impact:

Precision improved from ~14% base rate to ~22% at 50% recall (~1.5× lift)
~15% of all subscribers captured by targeting top 10% scored customers
Leakage-free modeling ensuring deployable, non-inflated results

5. AI-Driven Fraud Case Summarizer (Risk Explainability)

An AI-assisted fraud analytics system that combines model risk outputs, SHAP-based explainability, and GenAI narratives to transform raw transaction data and fraud signals into structured, analyst-ready case summaries—supporting faster, more transparent fraud review without automating final decisions. Read More →

Fraud Analytics Risk Modeling Explainable AI SHAP NLP GenAI LLMs Prompt Engineering Streamlit BFSI Decision Support

Key Impact:

Consolidates risk scores, explainability signals, and narratives into a single case view
Reduces analyst effort spent interpreting model outputs
Standardizes fraud case documentation with structured summaries

Links:

6. Vision-to-Language Engine

A computer vision and sequence learning system that converts raw visual signals into coherent natural language descriptions, combining deep CNN-based visual understanding with LSTM-based language modeling to bridge perception and human-readable interpretation. Read More →

Computer Vision Deep Learning CNN LSTM Vision-Language Models NLP Sequence Modeling Python TensorFlow Research-Grade

Key Impact:

Automated translation of visual content into descriptive language
Semantic alignment between visual features and linguistic tokens
Foundation for assistive, search, and content intelligence systems