• Designed and implemented a graph-based algorithm to find the optimal return path for an empty leg, resulting in a 10% increase in profit.• Developing AI-based Fleet Management and Scheduling software using Nextjs with T3.• Developed Charter Broker using React Native and OpenAI to effective book charter flights using Natural Language• Architecture effective state management using Apollo and GraphQL resulting in a 15% reduction in overall load time.• Implemented UI/UX enhancements, increasing user ratings by 20%.• Introduced Cypress and Detox testing integrated into CI to improve product quality for the web and mobile app respectively.

Persona is building a banking solution for communities to effectively pool funds and use them quickly.My work at Persona:• Successfully migrated Node.js Express to Nest.js on Google Cloud Platform (GCP), boosting overall throughput by 10%.• Architected and deployed a secure financial infrastructure using Stripe APIs, Node.js, Express.js, and TypeScript on Heroku.• Engineered transaction flows for mobile and web, elevating user satisfaction from 3.5/5 to 4/5 through a streamlined UI.• Enhanced reliability of notifications by implementing GCP Cloud Functions, resulting in a remarkable 40% reduction in downtime.• Led migration from React Context API to Redux, optimizing data organization and performance, reducing application start time from 5 seconds to 2 seconds.• Developed custom hooks and memoized functions using React Hooks, achieving significant performance improvements, including a 20% reduction in rendering time.• Improved chat module performance in React Native by implementing native Android features with Kotlin, resulting in faster response times and reduced lag time.

• Collaborated with the Department of Science and Technology (India) and Elsevier Research Journal to solve a problem of predicting the confidence score of the reviewer.• Created baseline Machine Learning models and Deep Learning models to test the robustness of our developed model using NLTK, Sklearn, and Tensorflow• Designed the model using BERT/SciBERT Embeddings with a hybrid CNN and LSTM network to increase the R2 Score by 5%• Analyzed the dataset to find a high negative correlation (p = -0.65) between the Conviction and Hedge Score• Designed a model which employed Attention on the LSTM network using features extracted from BERT/SciBERT, combined the attention vectors with Hedge Score and Conviction, resulting in to increase in R2 Score by 15% from the baseline model

• Developed a Cross Platform Augmented Reality based E Commerce App to handle request of over thousand request per hour in React Native and NodeJS• Planned the flow of app (React Navigation) to reduce efforts to user to find any item, additionally maintained all the APIs with Redux• Replicated all the design components to app components with less than 5% of difference.• Deployed the app on the PlayStore and App Store.