Tuning and behavior molding of Soopra's AI Personas.

Focusing on minimizing hallucinations and generating as close to accurate responses as possible.

Worked on Conversational modeling where AI will ONLY teach from a set of curricula defined by a professor/teacher/expert.

• Designed and implemented RAG Pipelines by experimenting with vector databases, resulting in 20% faster data ingestion.

• Developed a data Retrieval Engine utilizing a decay function for reverse chronological data retrieval with high embedding.

• Engineered RAG-based chat systems, dynamically enhancing conversation flow and increasing user engagement by 25%.

• Developed scalable ETL pipelines for real-time Twitter data extraction, preprocessing, and ingestion.

• Designed and optimized data ingestion pipelines, achieving a 15% reduction in data loading time into data stores.

• Developed conversational scripts optimizing user interactions and enhancing data extraction personalization, resulting in a 30% boost in user satisfaction and a 20% reduction in onboarding time.

• Developed a Django-based project management web app, increasing team efficiency by 30% and improving client updates.

• Implemented billing estimates and site inspection report features, enhancing project monitoring and documentation accuracy.

• Deployed the project management platform on Heroku with RDS, ensuring seamless data management and accessibility

• Consolidated multiple tools into a single platform for architectural and construction project management, providing efficient record maintenance and estimation.

Additionally, contributed to the project by working collaboratively with the team and using Heroku and RDS for seamless deployment and hosting. Accomplishments include streamlining project management, improving client communication, and enhancing team productivity.

• Programmed an SOS button for MyresQR, an emergency response platform, which sends alerts to the user's emergency contacts in case of an accident.

• Implemented IVR call flows using Exotel API to connect victims or respondents to emergency contacts quickly.

• Built services using Python in Django Web Framework, which enhanced the platform's capabilities and reduced emergency response times.

• Implemented and experimented with collaborative filtering algorithms for user recommendations on Netflix Movie Database.

• Formulated and developed a novel collaborative filtering algorithm to improve user recommendations.

• Presented findings to the research team and successfully integrated recommendations into the organization's offerings.