Headline

Quantum Computing Intern @ QRACE | Ex-Summer Intern @IIT Guwahati | ML Lead @KIIT Robotics Society | ML Team Member @AISOC | 2x Hackathon Winner

About

I am a dedicated and versatile technologist with a strong foundation in C and Python. Currently, I am expanding my programming skills by learning Java. My interests lie in Artificial Intelligence and Machine Learning, fields where I possess a solid grounding and a desire to delve deeper in the future. I am deeply passionate about AI, machine learning, and their applications in cutting-edge technology, particularly in the development of advanced weaponry. Currently, I am focused on mastering deep learning, generative AI, and Java, as these technologies play a crucial role in building intelligent and autonomous systems. My goal is to integrate these skills to develop state-of-the-art weaponry and defense systems that enhance strategic capabilities. I am eager to collaborate with experts and organizations pushing the frontier of AI-driven military technology. Let's connect and shape the future of defense together.

Location

Bhubaneswar

AI stack

aistudio

anthropic apis

antigravity

claude code

claude

codex

Work experience

ML Team Lead at KIIT ROBOTICS SOCIETY, Jan 2026 - Present · 8 mos

Artificial Intelligence Intern at MedEvidences, Nov 2025 - Present · 10 mos, Remote

ML Team Member at AISoC, Mar 2025 - Present · 1 yr 6 mos, Bhubaneswar, Odisha, India

Research Intern at QRACE-QUANTUM RESEARCH AND CENTRE OF EXCELLENCE, Jul 2026 - Aug 2026 · 1 mo, Delhi, India

Summer Research Intern at Indian Institute of Technology, Guwahati, May 2025 - Jun 2025 · 2 mos, Guwahati, Assam, India, While implementing and analyzing the TinyCLIP framework, I recognized a limitation in the use of contrastive loss. Contrastive loss, although powerful for aligning image–text pairs, assumes all incorrect (negative) pairs are equally dissimilar, applying a uniform penalty. This uniformity can overlook **hard negatives**—non-matching pairs that are semantically close to the anchor—leading to insufficient separation in the learned embedding space. To address this, I replaced the contrastive loss with a triplet loss function, inspired by metric learning techniques used in face recognition (e.g., FaceNet).

ML Team Member at KIIT ROBOTICS SOCIETY, Nov 2024 - Feb 2026 · 1 yr 4 mos

Education

Bachelor of Technology - BTech, Computer Science, Kalinga Institute of Industrial Technology, Bhubaneswar

Doon International school

Public links

github: https://github.com/a34656