Nirnay Roy Email & Phone Number

Pune, Maharashtra, India

◦ Automatic Profile Segmentation: Developed and implemented an advanced NLP-based algorithm that successfully classified 250 million LinkedIn profiles by their seniority and job function department. Enabled targeted marketing campaigns with over 90% precision, resulting in improved customer engagement and conversionrates.◦ Customer Success: Collaborated closely with the Customer Success team, providing valuable sales insights and recommendations to enterprise clients. Leveraged… Show more ◦ Automatic Profile Segmentation: Developed and implemented an advanced NLP-based algorithm that successfully classified 250 million LinkedIn profiles by their seniority and job function department. Enabled targeted marketing campaigns with over 90% precision, resulting in improved customer engagement and conversionrates.◦ Customer Success: Collaborated closely with the Customer Success team, providing valuable sales insights and recommendations to enterprise clients. Leveraged data analytics to identify opportunities for growth, resulting inincreased customer satisfaction and retention.◦ Data Engineering and MLOps: Played a key role in establishing robust ETL (Extract, Transform, Load)processes, ensuring efficient data pipeline management and data quality leveraging tools such as Airflow, Azure Databricks, and CircleCI.Implemented monitoring systems to track the performance and health of ML models,ensuring their reliability and scalability. Show less

◦ Adverse Media Detection: Achieved a 95%+ accuracy rate in detecting fraudulent activities, enabling effective due diligence checks advanced NLP techniques such as sentiment analysis, named entity recognition, coreference resolution, and POS tagging.◦ Knowledge Graphs: Leveraged a robust knowledge graph of people and organizations using Neo4j. to extract relationships with adverse entities, contributing to improved risk assessment and decision-making.

Varanasi Area, India

◦ Attention Mechanism: Outperformed the best model submission for a Named Entity Recognition task ofsocial media tweets dataset at WNUT-2017 to include an attention layer for better performance◦ Language Modelling: Designed and trained a language model pre-trained on a larger corpus with Attentionfor Named Entity Recognition which improved the F1 score by 2.5%.