Kolkata, West Bengal, India
I am a Technology Futurist. I create the Technology Solution of Future. Working as AI Architect in Cognizant Technology Solutions - India Having 8+ years of experience as Solution Architect, AI/ML. Total 19 yrs. Cross Industry Experience. Research interest in Program Synthesis. Published multiple research papers in Machine Learning Awarded patent with USPTO on Automated Exploratory Data Analysis. Conduct tutorial talks at Conferences and Data Science group’s meetups in India. Designing Custom Algorithms for Automating using AIML. Designed system for Time series AutoML, Observational Causality, NLP. Working on building systems with Quantum ML, Reinforcement Learning, Knowledge Graph.
1. GPT4 based HR Interview BOT 2. GPT3 based customer service BOT 3. GPT3.5 + KG based Claim Handler virtual Assistant 4. Quantum Computing for Insurance Whitepaper 5. Generative AI training plan, trained 40 GenAI Experts. 6. Create framework for EU AI Act compliance. 7. Advanced Time series analysis and forecasting using physics based ML 8. Reinforcement based Causality 9. NLP use cases using transformers, explainability and uncertainty for debugging. 10. Active Learning for data labelling. 11. OR using Quantum Hardware 12. Knowledge Graph extraction from Text 13. GPT3 and CODEX evangelist. 14. Automated Scientific Writing 15. Semantic Search Engine 16. Text based Information Extraction. 17. Automated EDA - Patent with USPTO
Business wants to reduce manual effort by Scientists for Publication. Scientists spend around 2 months, doing literature review, analyzing the information and then positioning original work in context of the literature for publishing to a top tier Journal like Nature. The publication has to be written up to Journal standards. This process of Search – Extract – Generate can be assisted using NLP technologies, and the objective of the project is to build: 1. Literature Database a. Remove Anomaly Edges using Topological Data Analysis on Citation Network embedding and document embeddings. 2. A custom search engine, such that a. The scientist is assisted with search query refinement to only display the papers that will be used in bibliography, using Synonyms and Connected words, for keyword discovery. b. Rank each paper using a novel ranking score, since H Index is local metric. 3. Extract Knowledge Graph using Snorkel for Information Retrieval using Named Entity recognition and Keyword extraction as Inputs to the Heuristic rules. 4. Summarize Literature and generate each section of the publication individually, using respective source documents using Deep Learning NLG model – T5, Bottle Sum, GPT2 Role: Lead AI Designer Skills: Research, ML System Design. Responsibilities: 1. Mentoring three NLP/NLG experts by designing the system, and defining experiments. 2. Define and Design the system.
Cognizant Adaptive Data Foundation™ is the reference model for creating a cloud-first, AI-driven data ecosystem that enables organizations to generate maximum business value from their unique data assets. It helps organizations develop their new data and analytics foundations required for our digitally savvy consumer world. Link : https://www.cognizant.com/cognizant-digital-business/applied-ai-analytics/ai-data-foundation DataIQ refers to the maturity of business data intelligence. The higher the DataIQ score, the more the data enables the organization to drive the right business decisions. Our DataIQ analysis identifies the existing—or missing—data sets that are critical to achieving business objectives. The tool helps us score and value data assets’ relevance and intelligence, enabling us to focus our data engineering efforts appropriately to deliver desired business outcomes. As a client’s DataIQ increases, so does the organization’s ability to perform descriptive,diagnostic, predictive and prescriptive analysis by identifying relevant information, reducing noise and redundancy and understanding the scale of complexity in the dataset. Link : https://www.cognizant.com/Resources/cognizant-adaptive-data-foundation-offering-overview.pdf