Next Generation of Travel Planning: Unparalleled, Award-Winning, and All-Inclusive first-of-its-kind platform in the world.

Project: Smart Text ExtrapolatorProject Description: Supporting Bajaj Finserv initiatives for Artificial Intelligence Solutionsin the areas of data Predictive AI (Smart Extrapolator involving Text Data from DifferentFinance Application Sectors). This project involved developing a Full Stack Web Application fordata Collection & Extraction along with interfaces for Data Interpretation & Smart Inferences.The system handled an average of 2000 Unique Loan Descriptions for each Sector (Internal &External). Output Descriptions were handed at Plagiarism of <5 % on an average with 0-2 secsresponsiveness depending on the Required Character Length.Standout Experience: Worked on AWS, DASK & Distributed Systems to implement the DataExtraction, Preprocessing, Training & Post Processing Parallely on 3 Different Loads (5 Inhousemachines, 5 VMs , 7 EC2 instances with a common EFS.Client: Bajaj FinservRole: Independent Developer

A) Smart Resume Analyser: Supporting UN initiatives for Artificial Intelligence Solutions in the areas of Data Predictive AI (Smart Analyser involving Text Data from 10 Different Job Sectors). This project involved developing a Full Stack Web Architecture for Data Collection & Annotation along with interfaces for Data Interpretation & Smart Inferences. 1. I was Responsible for Scripting the complete Architecture, Model Training, Testing and Deploying the Deep Learning solution alongside managing multiple tracks from both onshore client and offshore team.2. The system handled an average of 10,000 Candidate profiles & 550 skills for each Job Sector. Skills were Predicted at Precision of 97.7% with 0-2 secs responsiveness depending on the Profile Character Length.Tools - Python, Tensorflow, Flask, Docker, ReactJS, CSS, HTML, SQL, Cloud Instances, Neo4j Graph DB, NGINXB) Fraud Detection and Liveness Monitoring for an existing Face Recognition Engine (R&D):1. My Contribution involved researching on publications (Depth Analysis, 3ddfa Face Models, Face Passcodes, Face Distortion) for a robust Industry Standard approach. We achieved 93.1 F1 Score on a single layer base architecture with 0.3 sec responsiveness.Tools - Python, Tensorflow, Pytorch, SK Learn, Keras, JS, Flask, Django, NgrokC) Text2Sql ChatbotDevelop a Chatbot Interface, that queries Custom Client side SQL DB for the solution to respond the user.1. My responsibilities included managing both onshore client and offshore team. Guided & researched on existing feasible Text2Sql Solutions. Modified to satisfy the Client needs. Helped develop the front end and back end architecture for the Chatbot. We used Microsoft RATSQL alongside BERT Embedding space as the solution.Tools - Python, Tensorflow, Flask, Docker, ReactJS, CSS, HTML, SQL, Cloud Instances