Machine Learning Engineer
Hyderabad, Telangana, India
* Implemented a Bulk Post Similarity algorithm using FastText word embeddings to identify and remove plagiarism in Telugu language posts by local journalists. The algorithm combined Cosine similarity and Jaccard similarity metrics, leading to a 35% reduction in operational workload.* Developed and deployed a REST API using Flask to extract summarized text from audio in YouTube videos. The API utilized the IndicAsr Machine learning model to monitor news happenings on YouTube efficiently.* Conducted sentiment analysis of tweets from high-profile celebrity users on Twitter and classified their tweets' priorities using the Zero-Shot classification model. Utilized this analysis to curate trending tweets and posts in-app, resulting in a 20% increase in reach.* Created and implemented a solution for Telugu newsgathering from YouTube. The system took YouTube URLs as input and generated concise and informative summaries of news content.* Utilized web scraping techniques (Beautifulsoup, Scrapy, feed parsers, requests) to crawl web pages from various websites, ensuring the availability of the latest news.* Developed solutions for lazy loading websites using Selenium, expanding the capacity to gather more news content for the Way2News app.* Implemented IP rotation and proxies to avoid fingerprint detection and enhance web scraping capabilities.* Proficiently worked with MongoDB, Redis, SQL, and Node.js for efficient database management and handling.These achievements and initiatives contributed significantly to the organization's success and advancement in the field of Telugu language journalism.