Hi, I'm Om Nankar

Data Scientist

Machine Learning Engineer

AI Practitioner

Production-focused Data Scientist with 2+ years of experience engineering scalable ETL pipelines and deployment-ready ML systems. Specialized in automating complex workflows using Agentic AI and distributed computing.

View My Work → View My Resume →

Get in Touch

0.9847

∇loss

epoch:42

acc:97%

Experience

Where I've worked and the impact I've made

Data Science Intern

Chista

Aug 2025 – Present

Key Achievements:

Engineered an autonomous ingestion pipeline using CrewAI (Agentic LLMs) to parse unstructured healthcare data, reducing client onboarding time by 40%
Built a "Golden Layer" validation framework in PostgreSQL, implementing rigorous data quality checks that eliminated consistency errors across complex formulary datasets
Deployed a Traceability Audit Tool (using FastAPI/Streamlit) allowing stakeholders to visualize model lineage and source data

Python CrewAI PostgreSQL FastAPI Streamlit

Data Scientist

Wolters Kluwer

Jan 2023 – Aug 2024

Key Achievements:

Developed an automated pricing engine in Python, replacing a legacy 7-day manual process with a 1-click execution pipeline that reduced operational overhead by 70%
Designed scalable ETL workflows to aggregate attrition and revenue metrics, feeding real-time Power BI dashboards used by the Finance Centre of Excellence
Optimized SQL queries for high-volume financial reporting, restructuring data models to improve report generation speed for 10+ critical business metrics

Python SQL Power BI ETL

Research Intern (Computer Vision)

Symbiosis Centre for Applied AI

Jun 2021 – Aug 2023

Key Achievements:

Developed deep learning pipelines (PyTorch) for medical imaging (MRI) and public safety object detection, resulting in 7 peer-reviewed publications
Benchmarked detection architectures, optimizing model inference for edge cases in diverse datasets (dementia, Lyme disease)

Python PyTorch Computer Vision

Projects

Featured work and side projects

⚡

Distributed Big Data Recommendation Engine

Engineered a scalable ETL pipeline handling massive unstructured datasets using PySpark on a Hadoop cluster. Built a Hybrid Recommendation System combining collaborative filtering and content-based approaches for high-precision user preference prediction.

PySpark Hadoop Python

🔍

Real-Time Anomaly Detection System

Architected a streaming data pipeline using Apache Kafka for ingestion and Flask for serving, enabling sub-second latency anomaly detection. Deployed unsupervised learning models (Isolation Forests) via REST APIs, containerized with Docker.

Kafka Flask Docker REST APIs

🌍

CO₂ Forecasting Engine

Developed an end-to-end CO₂ emissions forecasting pipeline by integrating unsupervised learning and predictive modeling techniques. Applied K-Means and Agglomerative Clustering to segment regions based on historical emission patterns and climate indicators. Built and compared time-series and machine learning models, including Holt-Winters, ARIMA, Linear Regression, SVR, and Random Forest, evaluating performance using MAE and RMSE to identify the most robust approach and generate cluster-specific forecasts for climate trend analysis.

Python Scikit-learn ARIMA Clustering

View All Projects →