Vidhu

• Conducted a 12-week research project on Deep Q-Value learning in stochastic environments, focusing on stability and performance under noise and uncertainty.
• Designed an Adaptive Noise Injection mechanism to improve exploration and training stability in stochastic Deep Q-learning.
• Achieved more stable and higher Mean Return, consistently outperforming standard Deep Q-learning methods.

• Worked at RCMA-DRDO, Lucknow on Stall Warning & Protection and Brake Management Systems for the SARAS aircraft, gaining exposure to safety-critical aerospace and embedded software systems.
• Contributed to technical brainstorming and concept development during project discussions.
• Applied structured problem-solving to address unexpected technical challenges efficiently.

• Worked on Satellite Image Dehazing Using Deep Learning.

• Explored Fourier Transform and Bilateral Filtering for image clarity enhancement.

• Completed a technical internship at NEXA Defence (I) Pvt. Ltd. working on “Special Power Supplies for Aerospace Applications”, with emphasis on software-centric analysis of defence-grade embedded power systems and their operational constraints.
• Gained hands-on exposure to embedded systems programming concepts, including firmware workflows, sensor interfacing, signal conditioning, and real-time constraints relevant to aerospace and defence applications.

• Worked on the Design and Development part of a fully responsive webpage for a particular section of Desi QnA.
• Worked on User Verification/Authentication + Engineering challenges of detecting and removing the fake bots and posts.
• Integrated Recaptcha with the website - Google Layer of Protection for websites.