• Developed an automated system for large-scale scraping of industry-related news from search engines.• Maintained and expanded the news processing funnel, enhancing fault tolerance and reducing human involvement in the news process and selection.• Fine-tuned and prompt-engineered large language models (LLMs) for tasks including news summarization, market reports, and retrieval-augmented generation (RAG) based customer support systems.• Developed APIs for various ML-intensive tasks,… Show more • Developed an automated system for large-scale scraping of industry-related news from search engines.• Maintained and expanded the news processing funnel, enhancing fault tolerance and reducing human involvement in the news process and selection.• Fine-tuned and prompt-engineered large language models (LLMs) for tasks including news summarization, market reports, and retrieval-augmented generation (RAG) based customer support systems.• Developed APIs for various ML-intensive tasks, such as listing image understanding, lead responseautomation, and agent similarity detection.• Implemented a hierarchical location database for North America, detailing from country to neighborhood levels. Show less

• Evaluated various data sources (e.g., installed apps, search histories, page views) for their potential usein recommendations.• Applied clustering methods at scale to segment apps and users for data insights.• Implemented, tested, and compared collaborative filtering methods, including neighbor-based routinesand matrix factorization techniques (e.g., SVD variants, weighted ALS, Deep Factorization).• Developed a hybrid, multi-objective recommender system, increasing the… Show more • Evaluated various data sources (e.g., installed apps, search histories, page views) for their potential usein recommendations.• Applied clustering methods at scale to segment apps and users for data insights.• Implemented, tested, and compared collaborative filtering methods, including neighbor-based routinesand matrix factorization techniques (e.g., SVD variants, weighted ALS, Deep Factorization).• Developed a hybrid, multi-objective recommender system, increasing the overall click-through rate by 47%.• Collaborated with DevOps and Data Engineers to scale solutions using technologies such as Spark. Show less

• Developed the keyword generation component of the Search Ads system, enhancing Ad campaignrelevance and increasing click-through rates by 32%.• Created an automated web scraping module to extract pertinent data and related keywords for Ad campaigns.• Regularly monitored the Search Ads system’s performance, consulting with business intelligencespecialists to boost click-through rates.• Implemented a custom spell checker to map misspelled queries to existing keywords… Show more • Developed the keyword generation component of the Search Ads system, enhancing Ad campaignrelevance and increasing click-through rates by 32%.• Created an automated web scraping module to extract pertinent data and related keywords for Ad campaigns.• Regularly monitored the Search Ads system’s performance, consulting with business intelligencespecialists to boost click-through rates.• Implemented a custom spell checker to map misspelled queries to existing keywords, increasing thematch rate by 9%.• Classified millions of Instagram comments to identify sponsored and offensive messages. Show less