Experience

Role Overview

Working as a Data Scientist in WPP's advanced analytics and media intelligence arm, focusing on large-scale audience intelligence, ML-powered decision systems, and LLM-driven analytics workflows.

Key Responsibilities

• Architected the Audience Creation module of the Audience Insights App using Python Dash, enabling audience generation across 4.5B+ demographic, financial, and transactional records
• Optimized Snowflake SQL pipelines for audience creation, feature engineering, and profiling workflows at scale
• Trained and deployed ML models to score audience propensities for global brands
• Delivered Fusion-as-a-Service pipelines that map survey data onto national populations using donor-recipient feature fusion
• Collaborated closely with product, engineering, and analytics stakeholders to productionize ML systems with real business impact

Technologies

Python, SQL, Snowflake, Machine Learning, Feature Engineering, Propensity Modeling, LLMs, RAG, LangChain, Dash, FastAPI

Role Overview

Led end-to-end data science and ML initiatives in the real estate and preservation domain, spanning computer vision, optimization, and LLM-based insight extraction.

Key Responsibilities

• Built an automated property image annotation system processing 10,000+ images, improving preservation fund allocation efficiency by 20%
• Designed and trained CNN-based image classification models and clustering pipelines to categorize properties by condition and architectural features
• Developed an optimization algorithm using linear programming to allocate a $1M preservation fund, maximizing impact under cost constraints
• Built LLM-powered insight extraction pipelines to summarize unstructured real estate documents, improving reporting quality by 35%
• Implemented RAG-based systems combining structured market data with unstructured property descriptions for trend analysis
• Developed interactive dashboards and mobile interfaces to surface insights to non-technical stakeholders

Technologies

Python, NumPy, SciPy, TensorFlow, OpenCV, CNNs, LLMs, LangChain, Hugging Face, Flask, React.js, React Native, PostgreSQL, Elasticsearch, AWS

Role Overview

Worked as a Software Engineer at the intersection of ML, full-stack systems, and cloud infrastructure, contributing to healthcare AI, recommendation systems, and large-scale web platforms.

Key Responsibilities

• Fine-tuned BERT-based models for medical document summarization, reducing processing time by 80% across large healthcare datasets
• Built NER pipelines to extract clinical entities from unstructured medical documents, enabling faster decision-making
• Developed a hybrid recommendation system (collaborative + content-based filtering) serving 50K+ users, improving recommendation accuracy by 30%
• Built and deployed full-stack applications using Flask, React.js, and cloud services on AWS and Azure
• Led website migration initiatives using modern web frameworks, cutting page load times by 50%

Technologies

Python, C++, JavaScript, BERT, NER, Recommendation Systems, Flask, React.js, AWS, Azure, SQL, Data Pipelines