Welcome! I’m a data scientist passionate about solving real-world problems using machine learning, statistical modeling, and generative AI.
My work spans churn prediction, customer analytics, document automation, and intelligent agents powered by LLMs and multimodal pipelines.
I specialize in building end-to-end solutions — from data exploration and modeling to deployment and business integration — with a strong focus on explainability, automation, and performance.
- Churn prediction, time-to-event modeling, CLV forecasting
- Uplift modeling using causal inference for decision optimization
- SHAP- and chi²-based feature explanations for transparent ML
- Exploratory Data Analysis (EDA) and automated insights
- Funnel analysis and behavioral segmentation
- Custom KPI dashboards and data storytelling
- HR policy Q&A bots, meeting summarizers, email writing agents
- Retrieval-Augmented Generation (RAG) using vector stores
- Excel-integrated LLM pipelines for intelligent spreadsheet automation
- PDF image extraction and visual summarization
- Image-based search and generative prompting for creative workflows
- Slide generation from Excel using LLMs and layout intelligence
A full-stack engine for customer churn prediction, CLV forecasting, and uplift scoring.
- Applied tree-based models and survival analysis
- Used SHAP for explainability and segment-level reasoning
- Developed custom confidence metrics for predictions
An Excel-to-slide LLM pipeline that transforms raw data into polished presentations.
- Annotated visuals, optimized text, and contextual narratives
- Supports financial and sales reports
- Visual layout and tone control with minimal manual input
Tracks project progress across transcripts and documents.
- Summarizes decisions and tracks incomplete actions
- Parses PDFs and transcripts to update project status
- Outputs updated summary PDFs with discussion insights
A multimodal pipeline for visual parsing and reasoning.
- Extracts images from PDFs using OCR + vision models
- Enables image-based classification, visual search, and slide augmentation
Personalized writing agent for enterprise emails and docs.
- Detects recipient context using embeddings
- Adapts tone, structure, and language based on goal
- Suggests persuasive, clear edits and content adjustments
- Sales Funnel Analysis: Phase-wise tracking (e.g., initiation → conversion) with drop-off root cause detection
- Retention Reporting: Automated cohort dashboards with churn, survival, and retention curves
- Keyword Signal Analyzer: Chi²-based keyword impact analysis for CRM and support tickets
- Temporal Drift Detection: Detects drift in labels and feature distributions over time
- Behavioral Segmentation: Clustering + supervised labeling to derive actionable customer groups
Python · Pandas · NumPy · Scikit-learn · XGBoost · PyTorch · LangChain · OpenAI API · SQL · Excel
Matplotlib · Seaborn · Plotly · Streamlit · Power BI
FastAPI · Docker · Azure · AWS · Git
FAISS · Chroma · Pinecone · Sentence Transformers
I enjoy working at the intersection of data, creativity, and decision-making.
Open to collaborations, consulting, and LLM use cases that bring AI to real-world business workflows.
Thanks for visiting my portfolio! 🌟
Let’s build something meaningful together.