Khush Patel Email & Phone Number

United States

United States

– Worked with Prof. Kursuncu at GSU and Prof. Ponnurangam on multimodal hateful memes detection.– Led creation of MMBT model employing deep neural networks, NLP and knowledge graph techniques to scrutinize both visual and textual components within memes. Achieved 81% accuracy.

Hyderabad, Telangana, India

- Collaborated with Prof. Karthik Vaidyanathan to develop adaptive software systems using reinforcement learning methodologies to reduce server cost and enhance user experience.– Automated server adjustments and dimmer value changes using Q-Learning and Proximal Policy Optimization. Current model surpasses existing baselines.

– Developed and deployed a scalable Zoom-meeting recording portal on Oracle Cloud using TypeScript– Spearheaded integration of speech-to-text model, achieving a 95% accuracy in meeting transcripts.– Utilized text summarizer LLM - BRIO to generate concise meeting summaries, reducing review time by 70%.

Dubai, United Arab Emirates

- Worked on smart contracts to fractionalize real estate using ERC20 and ERC721 tokens.- Built a proof-of-concept full-stack interface to interact with the smart contract. - Explored ERC1400 tokens to increase security.

Hyderadabad

Worked under Prof. Ashok Das on the application of blockchain in the Internet of Drones(IoD) environment. Used CBOR Object Signing and Encryption to allow authentication and access control between drones.

Remote

Automating Visual Merchandising by creating dynamic forms to take input from owners and running ML algorithms to check if store design is complying with requirements.Using MERN stack, Image Similarity Algorithms, Object Detection Algorithms

Remote

'Smart Dividers'- A unique solution to traffic and tailgating which would revolutionise travel by creating a road system in which moving road dividers, controlled by ML algorithms, allow for smart land allocation based on traffic density on either side of the road.• Led 8-month research and design thinking period with 4-person team to discover causes of traffic in Dubai.• Built virtual prototype of AI-driven dividers using Figma to divide roads according to density of traffic as a… Show more 'Smart Dividers'- A unique solution to traffic and tailgating which would revolutionise travel by creating a road system in which moving road dividers, controlled by ML algorithms, allow for smart land allocation based on traffic density on either side of the road.• Led 8-month research and design thinking period with 4-person team to discover causes of traffic in Dubai.• Built virtual prototype of AI-driven dividers using Figma to divide roads according to density of traffic as a solution to the problem identified through research.• Pitched idea to NUS and NTU university professors in Singapore and the Road and Transport Authority in Dubai.• Complied a 65-page pitch deck and 2 progress reports, assessing progress in line with design thinking objectives.• Received recognition for Best Presentation and Research Methodology out of 15 design projects. Show less

Master'S Degree, Computer Science

Bachelor Of Technology - Btech, Computer Science, 8.76 Cgpa

Activities and Societies: Merit List Awardee