Romal Sehgal

I am a data science analyst with 4+ years of experience applying data-driven solutions across technology and advertising sectors. I specialize in analyzing large-scale datasets, building machine learning models, and translating insights into strategic actions.

At WPP Media I led the development of a cookie-less synthetic audience solution using clustering, classification, and large language models, and implemented a stacked ensemble framework to optimize programmatic display campaigns. I developed interactive dashboards for real-time campaign monitoring and integrated MLflow to improve model lifecycle management, accelerating deployment efficiency by 20-25%.

Earlier at Honeycomb.ai I contributed to an NLP pipeline, led feature engineering with Word2Vec and GloVe embeddings, and collaborated with engineers to productionize models that improved prediction performance. I routinely conduct rigorous A/B and post-campaign hypothesis testing to measure impact against key metrics like impressions, clicks, conversions, and CPA.

I am passionate about using data to drive efficiency, improve user experiences, and support product and business growth. I collaborate closely with cross-functional stakeholders to align data science initiatives with business goals and ensure reproducibility, scalability, and measurable outcomes.