• Developing hardware in the loop test plans, resulting in more reliable system operations and testing procedures.

• Directed the development of wireless connectivity solutions between IC tablets and PCs over ROS, significantly improving communication and data transfer efficiency.

• Established and refined data collection protocols for NDI sensors, substantially increasing the precision and reliability of our data inputs.

• Designed and implemented a local Apt-Cache for Mender and pre-commits, optimizing development workflows and speeding up deployment cycles.

• Under Prof. Michael Everett, developing and optimizing algorithms that leverage reachability analysis with verification in the loop to efficiently learn neural network controllers for autonomous systems, ensuring safety constraints are met while guiding the system towards desired goal states.

• Exploring the combination of neural network-based control approaches with formal methods such as reachability analysis,, to synthesize provably safe and high-performance controllers for complex systems operating in uncertain environments, while maintaining computational tractability.

• Led the outreach team at GDSC Northeastern, organizing over 10 tech events in 5 months. Handled everything from planning to finding and talking to speakers, and working with other clubs to get everything ready on time.

• Took charge of setting up events, from coming up with ideas to making them happen. This involved choosing topics, reaching out to experts to speak, and coordinating with other clubs for bigger and better events.

• Led the development of the company's flagship product MVP with React.js and JavaScript boosting early adopter engagement by 40%.

• Enhanced page load efficiency by 30% through strategic implementation of code splitting and lazy loading techniques in Next.js.

• Integrated Google Analytics with BigQuery for real-time user interaction resulting in accelerated bottleneck identification by 50% significantly enhancing overall user experience.

• Improved API efficiency with caching and better data practices, cutting server load and boosting response times for a smoother user experience.

• Under Prof. Bonny Banerjee, preprocessed the IEMOCAP (approx. 12 hours) and RAVDESS (approx. 24 hours) datasets by removing silences through energy-based Voice Activity Detection (VAD) and normalizing volume using global mean and variance normalization, improving data quality for analysis.

• Applied a proprietary university-developed model to the dataset, leading to a significant increase in the accuracy of emotional output prediction in speech.