Snehal Varbhe

I’m a data science and analytics professional focused on turning large, messy datasets into operational impact. At the Centre for Development of Telematics, I work with telecom records at massive scale—analyzing 5M+ blocked and 3.1M traced mobile records to uncover region-wise theft trends and automate reporting.

I architected a geo-spatial hotspot detection system that identifies high-risk theft zones with around 85% accuracy, enabling proactive monitoring across districts. I also built an IMEI intelligence system on 5M+ lakh records, detecting 15% duplicate and suspicious device patterns and uncovering hidden fraud links using graph-based analysis.

To improve responsiveness, I analyzed 5M+ blocked devices’ theft complaint data, revealing 30% longer reporting delays in specific regions and delivering behavioral insights to improve response efficiency. I built a Field Support Complaint Analytics Dashboard (50K+ tickets) using SQL/Python, reducing average resolution time by 30% and highlighting top recurring defect patterns.

Earlier, as a Data Science Intern at Toshiba Software (India) Pvt. Ltd., I constructed deep learning pipelines for haze removal and precise segmentation. I delivered strong model performance using CycleGAN + U-Net (94.09% accuracy, 95% F1-score) and a cascaded U-Net design, while designing end-to-end solutions with practical computational efficiency.