Priyam Gupta Email & Phone Number

New York, NY, US

Ransomware Detection using ML in VMDK files.

Seattle, Washington, United States

• Under professor Tim Althoff, Working on enhancing the performance of large language models (LLMs) with a focus on time series.
• Developing a context-rich and well-vectorized multimodal dataset by writing scraping pipelines from scratch using Python to extract data from various sources, including research papers, news, etc

San Diego, California, United States

• Designed and implemented a GenAI generation pipeline using RAG in Azure, enabling consultants to respond to client actuary and PBM proposals faster.
• Engineered SQL data pipelines for a pharmacy recommendation system, leveraging Washington State pharmacy claims data. Used clever feature engineering techniques, to aim to aid consumers in making more informed pharmacy.

• I played a key role in the research and successful pilot implementation of Mettl's SpeechX suite product. This product, designed to identify Mother Tongue Influence in spoken English, generated millions of dollars in.
• Our team developed innovative hybrid systems and testing methodologies that significantly streamlined the process of identifying candidates with pronounced accents. This advancement was particularly beneficial for.

Worked on developing an efficient real-time cheating detection pipeline using principles of face detection, gaze estimation, and outlier detection. This enhanced the cheating detection capabilities of the Mettl Suite and reduced cheating cases by 10%.

• Developed a BERT-based language model for the accurate identification of Hindi-English code-mixed languages in tweets. By utilizing a large-scale, multilingual dataset, we enhanced model robustness and performance by.
• Leveraged NLP techniques, including subword tokenization (BPE and WordPiece) and RoBERTa optimization, to achieve state-of-the-art performance metrics in Hindi-English code-switched language identification.

• Researched and authored a review paper addressing the cosmological challenge of determining Hubble's constant. This paper enhanced understanding of key discrepancies in current measurements.
• Utilized telemetry data from various telescopes to verify and recalculate Hubble's constant using Type 1-A Supernovas and gravitational lensing.
• Queried large SQL databases for telemetry data and designed an efficient data cleaning pipeline. This optimized the preparation and utilization of large data sets for cosmological research.

Master Of Science - Ms, Data Science

Bachelor Of Technology - Btech, Computer Engineering