Pratik Verma Email & Phone Number

𝐀 𝐃𝐚𝐭𝐚 𝐬𝐜𝐢𝐞𝐧𝐭𝐢𝐬𝐭 𝐩𝐚𝐬𝐬𝐢𝐨𝐧𝐚𝐭𝐞 𝐚𝐛𝐨𝐮𝐭 𝐦𝐚𝐭𝐡𝐞𝐦𝐚𝐭𝐢𝐜𝐬 𝐚𝐧𝐝 𝐜𝐨𝐦𝐦𝐢𝐭𝐭𝐞𝐝 𝐭𝐨 𝐬𝐨𝐥𝐯𝐢𝐧𝐠 𝐫𝐞𝐚𝐥-𝐰𝐨𝐫𝐥𝐝 𝐩𝐫𝐨𝐛𝐥𝐞𝐦𝐬 𝐰𝐢𝐭𝐡 𝐌𝐞𝐭𝐫𝐢𝐜-𝐃𝐫𝐢𝐯𝐞𝐧 𝐦𝐢𝐧𝐝𝐬𝐞𝐭!With over 15 years of combined experience in Data Science, AI, and Engineering, I am a data enthusiast and innovator, continuously pushing the boundaries of what's possible with data and AI technologies. Currently leading a ~30 member team as the Senior Director of Data Science at Netradyne, I have architected and deployed cutting-edge AI solutions that drive business value and ensure road safety. My career journey features key roles at IBM, Samsung and top-tier research institutions, offering me rich expertise in machine learning, generative AI (GenAI), Internet of Things (IoT), and advanced predictive modeling.

Notable recent contributions include:* Leading the development and deployment of a virtual assistant using Large Language Models (LLM) and RAG, enabling dynamic data retrieval (NLP to SQL) and visualization that enhances user experience. 𝐃𝐫𝐢𝐯𝐢𝐧𝐠 𝐫𝐨𝐚𝐝𝐦𝐚𝐩 𝐟𝐨𝐫 𝐛𝐞𝐭𝐭𝐞𝐫 𝐮𝐬𝐞𝐫 𝐚𝐝𝐨𝐩𝐭𝐢𝐨𝐧. * Creating a real-time accident detection and classification algorithm combining sensor fusion with computer vision, achieving over 95% precision, recall even on very low g-force signatures. 𝐈𝐦𝐩𝐫𝐨𝐯𝐢𝐧𝐠 𝐝𝐫𝐢𝐯𝐞𝐫 𝐬𝐚𝐟𝐞𝐭𝐲 𝐨𝐧 𝐫𝐨𝐚𝐝. * Developing a propriteray driver scoring system (GreenZone) that balances advanced prediction with user-friendly insights for improved risk management. 𝐑𝐞𝐝𝐮𝐜𝐢𝐧𝐠 𝐚𝐜𝐜𝐢𝐝𝐞𝐧𝐭𝐬 𝐛𝐲 50%. * Designing an anomaly detection framework for IoT devices, leading to efficient predictive maintenance and reduced active monitoring. 𝐈𝐦𝐩𝐫𝐨𝐯𝐢𝐧𝐠 𝐨𝐩𝐞𝐫𝐚𝐭𝐢𝐨𝐧𝐚𝐥 𝐞𝐱𝐜𝐞𝐥𝐥𝐞𝐧𝐜𝐞 𝐮𝐬𝐢𝐧𝐠 𝐚𝐝𝐯𝐚𝐧𝐜𝐞𝐝 𝐭𝐞𝐜𝐡.

Listed skills include Neural Networks, Amazon Web Services, Linux, Algorithms, and 38 others.

• 

  Senior Director

  Netradyne

  Jan 2023 - Present

  - #GENAI Safety Manager Assistant: Architected and deployed an LLM-based virtual assistant that leverages advanced NLP techniques to SQL queries for comprehensive data retrieval. This system performs sophisticated Retrieval-Augmented Generation (RAG) tasks to dynamically fetch, summarize, and visualize actionable insights from telematics data at scale. It delivers a near-real-time, contextually relevant user experience, enhancing data discoverability and reducing the learning curve associated… Show more - #GENAI Safety Manager Assistant: Architected and deployed an LLM-based virtual assistant that leverages advanced NLP techniques to SQL queries for comprehensive data retrieval. This system performs sophisticated Retrieval-Augmented Generation (RAG) tasks to dynamically fetch, summarize, and visualize actionable insights from telematics data at scale. It delivers a near-real-time, contextually relevant user experience, enhancing data discoverability and reducing the learning curve associated with traditional dashboards.-Greenzone Score: Developed an innovative driver scoring system using sophisticated machine learning models while ensuring the results remain highly explainable to end users. This approach balances advanced predictive capabilities with clear, actionable insights, facilitating improved risk management and road safety.- Accident Classification: Created a real-time accident detection and classification algorithm that combines sensor fusion with computer vision data. The algorithm excels in distinguishing between normal driving maneuvers and genuine accidents, achieving precision and recall metrics exceeding 95%. It provides critical insights for immediate response and analysis based on subtle variations in g-forces and sensor signals.-Recommendation Systems: Devised a recommendation engine employing unsupervised learning techniques to analyze historical driver performance data. Performance metrics focuses on the accuracy of recommendations and their effectiveness in improving driver safety before incidents occur.-IoT Device Health Prediction: Designed an anomaly detection framework for IoT devices, utilizing both edge and cloud computing to monitor and predict device health. The algorithm effectively filters through extensive device data to identify critical failures versus transient issues. Key performance indicators include the accuracy of issue classification and the reduction in the number of devices requiring active monitoring. Show less