Skip to content
View pranavsp108's full-sized avatar

Highlights

  • Pro

Block or report pranavsp108

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
pranavsp108/README.md

Pranav Padmannavar

Product Data Science | Analytics Engineering | Experimentation | AI-Powered Analytics

LinkedIn | Email | GitHub | Kaggle


I am an MS in Analytics candidate at the University of Minnesota – Twin Cities, graduating in May 2026, focused on Product Data Science, Applied Machine Learning, Experimentation, and AI-powered analytics systems.

Most recently, as a Data Science & Optimization Consultant at Daikin Applied, I built predictive optimization workflows that identified $1.5M+ in annual manufacturing cost savings and reduced simulation runtime by ~90%. Previously, I worked for 2.5+ years at Tata Consultancy Services, supporting global retail data operations, reporting automation, SQL analytics, and cloud migration initiatives.

I am currently building portfolio projects around e-commerce product analytics, customer segmentation, purchase propensity modeling, A/B testing, recommender systems, and GenAI analytics copilots.


Core Focus Areas

  • Product Analytics: Funnel analysis, KPI design, cohort retention, customer segmentation, revenue analytics
  • Experimentation: A/B testing, treatment/control analysis, hypothesis testing, bootstrap confidence intervals
  • Machine Learning: Propensity modeling, imbalanced classification, recommender systems, feature engineering
  • GenAI Analytics: RAG, natural-language KPI Q&A, AI executive summaries, OpenAI API, FAISS, Streamlit
  • Analytics Engineering: BigQuery SQL, PySpark, Databricks, Tableau exports, data quality checks

Featured Projects

1. E-Commerce Product Analytics, Experimentation & AI Copilot Platform

GitHub Repo | Tableau Dashboard | AI Copilot App

Problem: E-commerce teams need to diagnose funnel drop-offs, understand customer behavior, prioritize high-value segments, predict future buyers, and evaluate campaign impact.

Solution: Built an end-to-end product analytics and experimentation platform using BigQuery SQL, Python, Tableau, Scikit-learn, Streamlit, OpenAI API, FAISS, and SentenceTransformers.

Highlights:

  • Analyzed 4.3M+ GA4 e-commerce events across 360K sessions and 270K users
  • Built Tableau dashboards for executive KPIs, funnel analysis, category performance, cohort retention, and customer segmentation
  • Developed a leakage-safe purchase propensity model with 0.847 ROC-AUC and 15.8x PR-AUC lift
  • Evaluated randomized email campaign lift using z-tests and bootstrap confidence intervals; Mens E-Mail drove +$0.77 revenue/user lift
  • Added an AI analytics copilot for natural-language KPI Q&A, metric documentation RAG, executive summaries, and experiment interpretation

Skills: Product Analytics, BigQuery, SQL, Python, Tableau, Scikit-learn, A/B Testing, RAG, OpenAI API, Streamlit, FAISS


2. Scalable Grocery Reorder Recommendation Engine

GitHub Repo

Problem: Online grocery platforms need to predict repeat purchases, personalize baskets, and identify cross-sell opportunities across millions of customer-product interactions.

Solution: Built a PySpark + Databricks pipeline on 30M+ Instacart order-product records, engineering leakage-safe user, product, user-product, recency, and reorder-behavior features.

Impact:

  • Trained Spark ML models for reorder prediction with 0.409 PR-AUC
  • Achieved ~4x lift over baseline and 58% Recall@10
  • Added FP-Growth market basket rules and customer segmentation for personalized retail strategies

Skills: PySpark, Databricks, Spark ML, Feature Engineering, Recommender Systems, Customer Segmentation


3. Forecasting Pipeline for Market Signal Analysis

GitHub Repo

Problem: Market data is noisy, volatile, and difficult to operationalize without repeatable forecasting and validation workflows.

Solution: Built an automated Python forecasting pipeline on 26+ years of global market index data, engineering lag, rolling-window, volatility, and momentum features.

Impact:

  • Compared tree-based, LSTM, and statistical baseline models
  • Designed repeatable validation workflows using Python, TensorFlow, AWS, and GitHub Actions
  • Structured the project as a reproducible forecasting and market-signal analysis pipeline

Skills: Python, Time Series, TensorFlow, Scikit-learn, AWS, GitHub Actions


Tech Stack

Programming & Analytics

Python  SQL  R  Git

Product Analytics & Visualization

Tableau  Power BI  Streamlit  Plotly  Google Analytics 4

Machine Learning & Experimentation

Scikit-learn  XGBoost  LightGBM  TensorFlow

Data Platforms & Cloud

BigQuery  Apache Spark  Databricks  AWS  Azure

GenAI & Applied AI

OpenAI  FAISS  SentenceTransformers  RAG


Current Direction

I am currently focused on roles in:

  • Product Data Science
  • Applied Machine Learning
  • Experimentation / Growth Analytics
  • Customer Analytics
  • Retail / E-Commerce Analytics
  • AI-powered analytics and decision-support systems

Pinned Loading

  1. ecommerce-product-analytics-experimentation ecommerce-product-analytics-experimentation Public

    End-to-end e-commerce product analytics platform with BigQuery SQL, Tableau dashboards, purchase propensity modeling, A/B testing, and an AI analytics copilot.

    Python

  2. instacart-reorder-recommendation instacart-reorder-recommendation Public

    End-to-end recommendation engine using PySpark to predict reorders, generate cross-sell insights, and segment customers (Instacart dataset)

    Jupyter Notebook

  3. stock-market-pipeline stock-market-pipeline Public

    An autonomous end-to-end MLOps pipeline for global market forecasting, featuring real-time data ingestion via Apache Kafka, automated LSTM retraining on AWS EC2/S3, and self-healing infrastructure …

    Python