Canada
I am an adept Machine Learning Engineer with a solid foundation in computer science, evidenced by my Bachelor of Technology from Amity University and a Master of Applied Computer Science from Concordia University. My professional journey spans roles in India and Canada, with a focus on developing innovative AI and ML solutions. At iConnect Montreal, I have contributed to projects like an automated human recognition system and a cutting-edge machine learning model for PDF information extraction. My skills extend to programming languages like Python, Java, and JavaScript, and tools such as AWS, Docker, and Kubernetes. I also have expertise in web development technologies like React and NodeJS. I am passionate about leveraging AI and ML to create impactful solutions, demonstrated by my diverse project portfolio that includes a driver's license information extraction system, an e-learning platform for children, and advanced facial recognition technology. I am keen on collaborating on projects that harness the transformative power of AI and machine learning, and I am always eager to connect with like-minded professionals and explore new technological frontiers.
•Developed and deployed an NLP-based solution on AWS to generate transcripts from recorded customer-agent calls stored in S3, enhancing customer support efficiency. •Implemented embeddings generated and Retrieval-Augmented Generation (RAG) to fetch relevant documents and generate suitable answers in response to agent queries, ranking using cosine similarity. •Conducted unit testing and validation of the entire pipeline to ensure robust and reliable performance in production.
• Applied Azure Text Analytics to process large-scale telecommunications data, generating detailed insights and actionable recommendations that improved client outcomes. • Developed and optimized a text embedding pipeline, enhancing anomaly detection and ensuring clarity and precision in AI-generated reports.
• Contributed to the integration and testing of advanced AI technologies including LangChain, GPT-3.5, GPT-4, Llama. • Engaged in the ongoing research and application of Prompt Engineering and Retrieval Augmented Generation (RAG) techniques for AIRIS development, a conversational AI chatbot, focusing on user engagement and content personalization. • Analyzed Retail-Dataset of current prototype and solve for various KPIs of using Apache Spark so to extract insights. • Analyzed and demonstrated that ValleX outperformed XTTS in generating authentic audio outputs, achieving a 20% improvement in voice synthesis quality developing data pipeline and analyzing data scripts along with AI analytical tools. • Finetuned XTTS model for multilingual audio synthesis and realignment using Sagemaker and S3, increasing accuracy by 60%.
• Developed an automated human recognition and tracking system that was used in a variety of applications such as surveillance, assisting in the maintenance of safety and security in a variety of settings using Reinforcement Learning (RLHF). • Developed an advanced machine learning model on Google Cloud Platform (GCP) that efficiently extracts key information from PDF documents, demonstrating my expertise in natural language processing and leveraging GCP's robust ML infrastructure. • Developed Software to predict Retail Sales using AWSageMaker using Spark. • Automated the client recommendation system by implementing A/B Testing (statistical modeling) to compare different versions choosing the one with increased efficiency by 42% and reducing time by 51%. • Developed a License detection and reading REST API in Node.js with Python and AWS S3 services.
Projects I have worked on and working on- 1) Improving Road Safety in Canada by Analyzing Vehicle Defects using Machine Learning Developed a machine learning model that can accurately predict the likelihood of a vehicle having a defect based on various features, such as the make and model of the vehicle, the type of defect, and the location of the defect. 2) Developing a Recommended System for Grocery Shopping in Berlin •Developed a recommendation engine that uses machine learning algorithms to suggest the best grocery stores based on user preferences and needs. •Built a user interface that allows customers to input their preferences and receive personalized recommendations for grocery stores in their area and evaluate the accuracy and effectiveness of the recommendation engine using metrics such as precision. 3) Credit card fraud detection system •Developed a deep learning model for classifying financial transactions (fraudulant, non fraudulant) with a high accuracy inorder to curb the multifaceted impacts of this illicit practice. •Conducted comprehensive online research to source datasets, subsequently evaluating them to identify the most suitable one for a credit card fraud detection system. I delivered an in-depth analysis with key insights and successfully deployed a Random Tree model using Streamlit. 4) Equijob - Using AI to Balance Bias in Job Descriptions • Developed LSTM Model and fine-tuning BERT to detect bias words, replacing them, and regenerating whole descriptions without any racial or gender-biased words. 5) UEFA EURO 2024 – Leveraging Machine Learning & Open Data Sets for Advanced Sports Analytic working on a project leveraging machine learning and open data sets to provide insights into team strategies, player performance, and the economic impact of UEFA EURO 2024 on host cities. This involves tackling data fragmentation, integrating diverse data types, employing advanced ML techniques for comprehensive sports and tourism analysis