Atharva Chaskar

Background

Experience

Data Research Assistant

University of Illinois, Springfield · Springfield, Illinois

Oct 2024 – May 2026

Conducted spatial segmentation analysis on 3,322 census block groups from the EPA Smart Location Database, applying K-means clustering to identify a Transit-Walkability Gap affecting 53% of the Chicago metropolitan area neighborhoods.
Built an end-to-end Python pipeline incorporating feature selection, z-score standardization, and one-way ANOVA validation, optimizing model stability by identifying and removing multicollinear variables.
Generated publication-quality spatial maps using GeoPandas and Matplotlib, visualizing cluster distributions across the Chicago–Naperville–Elgin CBSA.
Co-authored a peer-reviewed research paper published in the MWAIS 2026 Conference Proceedings and received the Best Research-in-Progress Paper Award at the 21st Annual Midwest Association for Information Systems Conference, Athens, Ohio.

Read Publication ↗ Live Tool: Commute-Honest Housing Score ↓

Associate Application Developer

Accenture · Data & Analytics · Pune, India

Apr 2023 – Jul 2024

Analyzed and transformed data from 5+ heterogeneous sources to support reliable reporting and business intelligence, improving data ingestion reliability by 15% through validation and profiling.
Developed and optimized SQL transformations within Azure Data Factory workflows, reducing query execution latency by 10% and producing KPI-ready datasets for executive reporting dashboards.
Built automated reporting and data-processing workflows that reduced recurring data-quality anomalies by approximately 15%, improving consistency of analytical outputs.
Collaborated with cross-functional technical and business teams during bi-weekly production releases, providing UAT support, participating in code reviews, and validating data and reporting changes.
Applied Git, CI/CD, and Agile/Scrum practices to support reliable deployment of analytics solutions while maintaining technical documentation and SOPs.

Education

M.S. Management Information Systems

University of Illinois Springfield

GPA 3.9 / 4.0 · May 2026 · Beta Gamma Sigma

B.E. Mechanical Engineering

JSPM's Rajarshi Shahu College of Engineering, Pune

May 2022

Work

Projects

Commute-Honest Housing Score Full City Map

Commute-Honest Housing Score Address Breakdown

Address-Level Scoring & Routing Interactive Live Demo ↗

Applied Analytics & Product

Commute-Honest Housing Score (CHS)

Productized peer-reviewed clustering research (MWAIS 2026) into a live, address-level scoring tool — replaced a single opaque "walkability" number with four transparent, weighted sub-scores covering commute, walkability, transit frequency, and infrastructure-vs-jobs mismatch.

581,834 Chicago addresses scored · Full city coverage

Productized peer-reviewed clustering research (MWAIS 2026, Best Research-in-Progress Paper) into a live, address-level scoring tool — replaced a single opaque "walkability" number with four transparent, weighted sub-scores covering commute, walkability, transit frequency, and infrastructure-vs-jobs mismatch.
Built an end-to-end pipeline (DuckDB + dbt Core, medallion architecture, 70/70 data quality tests) computing real transit+walk travel times via GTFS schedules and OpenStreetMap routing (r5py) — scaled to full-city coverage via a 9,525-point grid-based precompute architecture instead of 40+ hours of brute-force routing.
Validated composite scores against real Chicago geography with zero manual tuning (Loop/Near North Side highest, Hegewisch lowest) and replicated the original paper's 53% Transit-Walkability Gap finding almost exactly.

DuckDB dbt Core Python r5py scikit-learn FastAPI Leaflet Docker

Live Demo ↗ GitHub ↗ MWAIS Paper ↗

CTA Performance Analytics Platform Dashboard

25-Year Ridership Analytics Dashboard Interactive Live Dashboard ↗

Data Analytics

CTA Performance Analytics Platform

Analyzed 25 years of Chicago Transit Authority ridership data to uncover long-term mobility trends, post-pandemic recovery patterns, and differences between bus and rail usage.

11.1B rides · 25 years · 57% post-2020 decline

Analyzed 11.1 billion Chicago Transit Authority rides spanning 25 years using Python, Pandas, and SQL to identify long-term ridership patterns, including a 57% post-2020 decline and subsequent recovery trajectories.
Conducted year-over-year and mode-level analysis to evaluate bus versus rail ridership trends, translating large-scale transportation data into interpretable performance insights.
Built a Python ETL workflow to ingest Chicago Data Portal data into Snowflake and developed an interactive Plotly Dash application with four dynamic visualizations for exploring ridership trends and distribution.

Python Pandas Snowflake Plotly Dash REST API Render

Live Dashboard ↗ GitHub ↗

Olist Product Category Economics Dashboard

Looker Studio Executive & KPI Dashboards View Dashboards ↗

Business Intelligence & Analytics

Olist E-Commerce Analytics Pipeline

Analyzed approximately 1.5 million e-commerce transactions across nine datasets to evaluate revenue, delivery performance, seller activity, and product-category trends.

1.5M Transactions · $13.7M Revenue · 91.9% On-Time Delivery · 74 Categories

Analyzed approximately 1.5 million e-commerce transactions across nine datasets to evaluate revenue, delivery performance, seller activity, and product-category trends.
Developed three Looker Studio dashboards tracking key business KPIs, including $13.7M in platform revenue and a 91.9% on-time delivery rate across 74 product categories.
Used Python, Pandas, Snowflake, and dbt Core to transform raw transaction data into structured analytical datasets, including eight staging models and three data marts.

Python Snowflake dbt Core Looker Studio SQL

GitHub ↗

Chicago Crime Executive Overview Dashboard

Chicago Crime Geographic Hotspot Analysis

Tableau Operational & Hotspot Dashboards View Dashboards ↗

Data Analytics & Predictive Modeling

Chicago Crime Analytics & Predictive Modeling

Large-scale preprocessing and analysis of 750K+ Chicago crime records with Tableau dashboards and machine learning models for pattern detection.

Transformed 750K+ raw Chicago crime records into analytics-ready datasets using Python and Pandas, resolving missing values, type conversions, and structural inconsistencies across a large-scale public safety dataset
Designed 3 Tableau dashboards (executive overview, time/arrest insights, geographic hotspot analysis) visualizing patterns across 262K+ incidents for operational decision-support
Developed predictive analytical models including hierarchical clustering and Decision Tree classification achieving 88% accuracy for arrest outcome prediction, validated with 90% sub-sample stability testing

Python Pandas Tableau SAS Enterprise Miner Hierarchical Clustering