About
I build production ML systems end-to-end — data pipelines, features, models, and the deployment around them. Currently a Data Scientist at Tredence, working with enterprise clients on forecasting and customer decisioning. Recent work: - A ticket sales and revenue forecasting engine on Databricks for Live Entertainment Client. I own the primary and feature layers — 1000+ features engineered in PySpark — and contributed to cohorting, modelling, and deployment. - A Next Best Action framework for a telecom client: 12 propensity models across upsell, cross-sell, deep-sell, churn and retention, built in Kedro and deployed on AWS, with a rule-based arbitration engine that resolves competing model outputs into a single prioritized action per customer. - The backend for a RAG-based troubleshooting system — document retrieval, multi-step conversational workflows, and modular ingestion and vectorization pipelines. Cut query latency by ~30-40%. Outside client work, I write C++ for the same reason some people do crosswords. I've built a machine learning framework from scratch — backpropagation, gradient descent, and loss functions written by hand rather than imported — and a 3D graphics and physics engine running at ~60 FPS. Both are on my GitHub. Toolkit: Python, SQL, PySpark, Databricks, LightGBM, XGBoost, MLflow, AWS, Kedro, GitHub Actions, C++. IIT Jammu, B.Tech with a CS minor. Open to Data Scientist and ML Engineer roles — remote or Bengaluru. Reach me at hardikpatodi2209@gmail.com.
Work experience
Data Scientist at Tredence Inc., Aug 2025 - Present · 1 yr 3 mos, Bengaluru, Karnataka, India
Summer Intern at CSIR- National Chemical Laboratory, May 2024 - Jul 2024 · 3 mos, Pune, Maharashtra, India
