Bengaluru, Karnataka, India
Skilled in Python (Programming Language), C++, Machine Learning and Deep Learning. Strong research professional with a Master of Technology (M.Tech.) focused in Aerodynamics from INDIAN INSTITUTE OF TECHNOLOGY KANPUR.
• Developed and deployed a Question Answering System for FAQs and technical assessments based on previous Zoom chats, utilising Langchain and LLM models, to enhance 24/7 client support. • Building data sets for ML models to optimise test procedures using ArangoDB and Neo4j for efficient storage and retrieval of previous test records.
- Developed an AI extraction tool employing Table, Cell, and OCR detection models, deployed on AWS EKS. Utilised Faster RCNN and K-means, cutting document extraction time by 50% and AWS-Tesseract OCR costs by nearly 90%. Enhanced OCR performance with BERT. - ML models implemented in the Merging app, replacing pairwise comparison, accelerating data extraction, and merging. TF-IDF and LSA were used to identify contextual similarity for a 20% boost in data quality. - Developed NLP models with Transformer Decoder for efficient financial data extraction. BertSum and LDA were used, significantly reducing the average ingestion to extraction time from 90 to 15 minutes. - Built a Data Warehouse using AWS Glue for integration of the ETL pipeline from various sources into AWS Redshift, resulting in a 15% increase in client acquisition for the sales team. Enhanced document preprocessing, parallelised ingestion with AWS Batch and ECR, cutting the average document ingestion time from 18 to 4 hours for 10,000 documents, and reducing monthly AWS costs.
- Developed 10 DOE vehicle models. Performed simulations on it and used data modelling techniques to find the coefficients responsible for Tire Valorization. - Built an RR Model using Markov Chain Monte Carlo algorithm to predict the coefficient of rolling resistance of the tire. - Conducted a workshop of 5 days on Basic & Advanced Statistics Training for new hires of Michelin. - Conducted a training workshop of 2 days on Rolling Resistance and Fuel Consumption for 15 ARAI Engineers.
Developed a model employing the Markov Chain Monte Carlo algorithm to predict tire rolling resistance coefficients, impacting fuel consumption. Integrated seamlessly into an internal tool, empowering the sales team for a 10% increase in new customer acquisition.