Developed sophisticated AI models in Python3, leading to a remarkable $200K increase in system value and an annual value contribution of $1M, validated through rigorous financial metrics analysis
Applied conceptual understanding of algorithms, enhancing time series analysis accuracy by 25%
Engineered efficient data processing workflows, reducing processing time by 30% and implementing CI/CD pipelines that boosted deployment frequency by 50%
Created dynamic data visualization dashboards using Grafana, Python, and SQL, leading to a 50% improvement in monitoring accuracy
Actively pursued continuous learning, staying updated with the latest advancements in data science, algorithms, and Big Data frameworks such as Spark
Demonstrated exceptional teamwork, contributing to a 15% increase in team productivity and a 25% reduction in project completion time.
Data Science Engineer
CGI Inc
08.2017 - 09.2020
Enhanced fraud detection accuracy in insurance claims by 20% through the development of a machine learning ensemble model.
Rolled out a predictive model for credit risk assessment using Gradient Boosting and upgraded the loan approval process by 25%.
Built a natural language processing model for analyzing customer feedback in the banking sector using techniques such as sentiment analysis and topic modeling, resulting in a 30% improvement in customer satisfaction.
Conducted data cleaning and feature engineering, web scraping with Python.
Created traditional time series deep learning models and deployed to forecast sales with an intuitive approach, resulting in an 28% boost in accuracy in car loan sales for European market.
Utilized statistical analysis and data-driven insights from A/B tests to enhance customer engagement, conversion rates, and operational efficiency in the financial services sector.
Master Thesis- Incident Report classification according to Penal Codes
Trained summarization model on Indian legal documents, evaluating with ROUGE-2 score 82% , ROUGE-L score 37% improvement from pre-trained, implemented Named Entity Recognition model on legal documents to create database.
Composed vector database for Indian Penal code document using faiss db, langchain framework- by generating embeddings taking transformer model 'all-roberta-large-v1' from huggingface to compare with incident summary, evaluating with similarity search score.
Extended the system to utilize large language model from openAI to pick logical suggestion from vector DB using prompt engineering.
Certification
Natural Language Processing Specialization
Data Structures and Algorithms
Machine Learning Engineering for Production (MLOps) Specialization
Senior Account Manager at Magnibot Technology Solutions India Private LimitedSenior Account Manager at Magnibot Technology Solutions India Private Limited
Application/Senior Production Support Engineer at Magnibot Technology solutions Pvt LtdApplication/Senior Production Support Engineer at Magnibot Technology solutions Pvt Ltd