Anuj Saxena

• Leading the Data Science & Advanced Analytics team at IPG Health consisting of DS managers and Data Scientists.
• Building out multi-touch attribution capabilities to assist media teams optimize media mix. Approaches include non-parametric regression and timeseries, causal inference
• Building automated reporting solutions using Generative AI and LLM models. Approach includes using Langchain python interface, GPT4All, OpenAI modules.
• Built language solutions to find best search keywords to bid on for Paid search marketing opportunities by leveraging and combining disparate data sources, NLP embeddings, and LLM models.
• Built geographic marketing and media optimization tool in python to optimize geographic media heavy-ups.
• Managed a team of data scientists, set weekly expectations, bandwidth assessment, managed requirements and provided clarity on projects, being a mentor to the larger analytics group.

• Built audience segmentation and look-alike models using clustering algorithms in Python and Pyspark for strategic audience targeting initiatives. Approaches include KMeans, KModes and Hierarchical clustering, weighted Gower distance.
• Built geographic marketing and media optimization tool in python to optimize geographic media heavy-ups.

• Built Django APIs for B2B Healthcare Data Analytics platform.
• Created a classification model in Python to identify which Doctors are influencers in the medical space and potential followers for different clients to help with treatment advocacy and speaking engagements. Approach includes Logistic regression with L1 regularization, SMOTE (synthetic minority oversampling technique)

Developed lead generation metric using leading indicator website actions, generated automated weekly sales rep reports for 71 sales territories in US. This increased efficiency in the Sales team in reaching out to customers at the right time.

• Created pattern detection classification models to classify messy show titles
• Created automated Data pipelines of digital web data using R and Python Pandas
• Analyzed News articles based on user interaction and extracted relevant keywords from articles in Python
• Utilized click stream data to analyze user behavior on NBC News websites in Hadoop
• Used Natural Language Toolkit and Pandas to cleanse and analyze Facebook data for estimating Article reach on Social websites

• Created market based predictive models to estimate the sales of Clorox Disinfecting Wipes in R.
• Web scraped Amazon reviews of Clorox disinfecting wipes and its competitors to understand consumer sentiment

Built a Panel linear model to predict market-level sales soup, beer and insect spray using weather data, flu trends and other external factors. Acquired data from weather underground API to get weather variables using Python.