Research & industry experience

Member · Los Angeles, California

Developing Permissions That Learn: fixed user-defined boundaries, a decision gatekeeper, and approval-learning memory for AI agents. Evaluating interruptions, unauthorized actions, and permission creep in simulated AgentDojo tasks.

LLM Researcher · Chennai, India

Adapted LLMs and vision-language models for computational neuroscience and digital histology. Built pipelines for Nissl/H&E imagery, MRI-linked data, and biomedical annotations, covering alignment, fine-tuning, inference, and evaluation on NVIDIA DGX A100 infrastructure.

Vice Chairman & Lead Backend Developer · Chennai, India

Helped lead a 17-member team building the college’s production website. Architected Node.js/Express/MongoDB services and 50+ REST endpoints, with authentication, AWS deployment, CI/CD, monitoring, and security controls; established reviews, documentation, and successor mentoring.

Technical leadership & community service · Chennai, India

Progressed through three roles spanning community programs, volunteer coordination, and software development.

Software Developer Intern · Tamil Nadu, India

Developed ML stock-transfer prediction and verification, utensil-cleanliness classification, and inventory vision workflows for retail and Smart Café operations. Built end-to-end datasets and model pipelines; published clean/dirty glass classification, object detection, and Rangoli Detection datasets on Kaggle.

AI Software Engineer · Tamil Nadu, India

Developed EcoFloater’s onboard AI for aquatic-waste detection and navigation. Integrated LLM decision components, YOLOv8, DeepLabV3+, lightweight vision transformers, and sensor inputs on Jetson Nano; applied compression, quantization, and latency optimization for local mission execution.