Vodafone-Idea Limited | Big Data and Advanced Analytics Intern
January 2020 - July 2020 | Mumbai, India
At Vodafone Idea, I worked on analytics systems for customer sentiment, regional telecom trends, and churn-risk detection.
The day-to-day work was a mix of data engineering and applied ML. I was pulling noisy signals from social and telecom sources, turning them into structured features, and shipping outputs that business teams could actually use for retention and network-priority decisions.
What I Built
- Built a fully automated social media analytics dashboard with Flask and SQLite that combined Twitter and Google Trends data for telecom keywords across India. This gave teams a single place to track sentiment shifts and issue clusters by region.
- Implemented NLP outputs including unigrams, bigrams, supervised topic modeling, and manual topic modeling to surface recurring customer pain points. This helped separate random noise from patterns that repeatedly affected customer experience.
- Generated Google Trends keyword signals and mapped strong vs weak popularity zones across telecom operators. The output made regional performance gaps easier to compare and prioritize.
- Automated large-scale data mining and cleaning workflows over millions of records using Selenium and BeautifulSoup. This reduced manual data prep time and improved consistency in downstream analysis.
- Worked on churn analytics for the northeast region by identifying top-decile high-risk customer cohorts with confidence scoring. These cohorts were used to support retention-focused outreach and prioritization.
- Deployed ML models on GCP with Nginx and Gunicorn to make the analytics pipeline usable in a production-like setup, not just as offline notebooks.