Designed and implemented an ETL workflow to collect, clean, and integrate surname records from US Census and USPTO datasets using Python; performed exploratory data analysis and statistical validation to address data biases and reduce missing values by 30%. Architected a hierarchical, multi-stage demographic classification system for ethnicity detection from last names, starting with broad Census surname-based detection and refining predictions for Asian subregions via USPTO data, enabling fine-grained demographic insights with improved interpretability and scalability. Delivered a Flask-based web application with real-time capabilities and a user-friendly interface, achieving 99% precision in ethnicity prediction through character-level embeddings, LSTM, and NLP techniques.

ML Volunteer at C/test

December 1, 2024 - September 19, 2025

Transformed cross-functional marketing goals into an A/B testing assistant using GPT-4 to generate experiment variants (headlines, descriptions, CTAs) for email, ads, and landing pages; built multi-touch attribution models using GA4 and Meta Ads data to quantify ROI, improving targeting efficiency by 15%. Automated interpretation of test results with LLMs to summarize outcomes, surface key takeaways, and suggest actions, significantly reducing data team workload. Reduced experimentation cycle time by 30% by eliminating manual bottlenecks in variant design, summarization, and stakeholder reporting; created Tableau dashboards to visualize attribution insights for product stakeholders and KPI alignment.

Data Engineer Trainee at Shree Hari Info Solution

May 1, 2023 - September 19, 2025

Objective: develop ML-driven coaching-prioritization solutions by considering attributes affecting employee performance. Collaborated with HR leaders to define 5+ KPIs and engineered ETL pipelines to ingest, cleanse, and standardize 1M call-center records, improving training efficiency by 20% and reducing data preparation time by 50%. Created analytical hypotheses and validated them with statistical tests; built predictive models and deployed ETL pipelines on GCP to support decision making in employer performance. Projects included PCB and Wafer Defect Detection (integrating 50GB+ multi-source datasets with YOLOv8/CNN/ViT/Faster R-CNN achieving 96.4% precision) and Real-Time Bank Data Streaming Capture (Kafka/PySpark/AWS IoT/TimeStream) with end-to-end pipelines and dashboards (Tableau, Grafana) for defect and transaction monitoring.