About
I’m Kanav Atre, a Purdue senior in Data Science & Applied Statistics. I like to work across the stack, but have a strong inlcination in Machine Learning, Computer Vision and databases .
In my most recent internship, I built a real-time KPI pipeline with Neon, ClickHouse, and Postgres. I was also responsible for deploying and maintaining a data migration and archival infrastructure for fast, accurate and most up-to-date analytics. I also worked a bit on the frontend integration of supply-chain specific charts and built executive-facing dashboards and demos for prospective customers.
I also have a strong undergraduate research project in pedestrian detection and behavior analysis using large-scale image datasets to study changes in pedestrian patterns before and after the COVID-19 pandemic.

Skills & Technologies
A snapshot of what I reach for most often.
Languages
Frontend
Backend
Data
Infra
Tools
Experience
Jun 2025 — Aug 2025
Software Engineer Intern
ketteQ
Built real-time KPI pipelines across Neon, ClickHouse, and PostgreSQL; delivered interactive dashboards (bias charts, treemaps, stacked bars) used by analysts and leadership.
- Real-time analytics pipeline (Neon → ClickHouse/Postgres)
- Executive-facing KPI dashboards + demos for prospective customers
- Data migration + archival infrastructure to keep analytics fast
Aug 2023 — Jun 2024
Undergraduate Researcher
Purdue University
Pedestrian detection and behavior analysis project using large-scale image datasets to study changes in pedestrian patterns before and after the COVID-19 pandemic.
- Mentored by Prof. Carla Zoltswoski and Prof. Edward J. Delp
- Analyzed a dataset of 50,000+ images to study pre- and post-COVID pedestrian behavior
- Built models using KNNs, linear classifiers, and neural networks for feature extraction and accuracy improvement
Aug 2022 — Jan 2023
Data Science Intern
Renzoe Box
Worked on computer vision and data pipelines for RenzoeMatch, improving color classification accuracy and powering product recommendation systems.
- Improved the RenzoeMatch model by combining 3-point color matching with LAB-space feature extraction and brightness normalization for more reliable tone classification
- Engineered web-scraping pipelines using BeautifulSoup to collect structured product data from dynamic websites
- Built automated data ingestion workflows that directly powered RenzoeMatch’s recommendation engine