Aditya Krishna

Designed and developed a dashboard and Google Cloud Dataset for monitoring and alerting on video experience KPIs. Developed an automated emailing system for Engagement KPIs breakdown scenario, upgrading a manual process to an automated Looker dashboard. Designed a monitoring dashboard for negative reviews, incorporating sentiment analysis and clustering, which helped reduce negative reviews by 20 percent. Conducted business analysis on device metadata effects on engagement metrics to help the Ads team identify audience cohorts. Worked on Google Cloud Vertex AI for meme caption and image generation as part of a marketing studio team at Zee Hackathon. Worked on customer profiling to increase content discovery for new users.

Benchmarked spatial analysis methods (PostgreSQL vs GeoPandas), finding that PostgreSQL outperformed GeoPandas by 60 times. Designed and developed a Tableau dashboard for electric stations analysis in Florida state, aiding clients in key decision-making.

Improved ML model for sentiment classification, achieving 88.6% accuracy for tweets. Extended analysis to NER and Topic Modeling. Identified that only 7% of tweets were customer-related. Worked to identify anomalies in loan branches using various algorithms, identifying around 1.1 percent anomalies. Identified that around 28 percent of customers still used old methods for document retrieval, providing data for marketing decisions. Created and tested ML models for classifying tweets into categories (Spam, Grievance, Query, Feedback), achieving 82% accuracy using a two-stage model (SVM with GLOVE).