CS researcher at IIIT Hyderabad working on computer vision, machine learning & 3D vision — and currently a systems software intern at NVIDIA.
I work across computer vision, machine learning and 3D vision — building systems that understand scenes from images and point clouds, and reason about what they see.
I'm a 5th-year B.Tech (Hons.) + M.S. by Research student in Computer Science at IIIT Hyderabad, working as an undergraduate researcher at the Machine Learning Lab on open-vocabulary 3D scene understanding.
I like working across the stack — computer vision and machine learning, low-level systems, and sensor-driven hardware. From coursework to research, I enjoy taking an idea all the way to a working system, especially where perception meets the real world.
Implemented the SGLang backend in the RTVI-VLM microservice — part of NVIDIA's VSS (Video Search & Summarization) stack — and ran performance and accuracy testing across the pipeline.
Research on open-vocabulary 3D scene understanding and computer-vision systems — first author of a 2025 review on LLMs and 3D vision in Sensors (MDPI).
Managed a budget exceeding ₹30 Lakhs for the institute's flagship techno-cultural festival, from sponsorship through to settlement.
Organized campus-wide sporting events with 1000+ participants — coordinating teams, venues and schedules end to end.
Spearheaded robotics projects and mentored junior members of the Electronics & Robotics club.
Led the logistics team for artistic and cultural events across campus.
Oversaw daily operations of a campus mess serving 1000+ students.
Provided academic support and mentorship to incoming first-year students.
A dual-degree program focused on research in machine learning, computer vision and robotics — currently in the research phase at the Machine Learning Lab.
Senior secondary in Physics, Chemistry, Maths & Computer Science — 98% in Class XII, 94.5% in Class X.
A peer-reviewed review of integrating LLMs with 3D vision — methodologies, applications and open challenges for scene understanding and multimodal robotic decision-making.
Read the paper ↗Detection (ResNet, R-CNN, YOLO, ViT), segmentation (FCNN, U-Net), neural style transfer and saliency — including custom Vision Transformers.
MLP, CNN, HMM, GMM and RNN implementations across diverse tasks, reaching 90%+ accuracy through meticulous tuning and visualization.
Chroma keying, filtering, detection, segmentation, denoising and sharpening — plus an automated coin-counting pipeline.
An ESP32/Arduino wheelchair with 8 sensors monitoring user vitals at 95% accuracy, streaming to a ThingSpeak dashboard.
An NFS-inspired DFS in C: TCP sockets, multi-threaded concurrent operations and a server backup mechanism for fault tolerance.
A web app for music lovers: a Flask backend, a Jinja/JS frontend and a MySQL store powering playlists, search, charts and reviews.
Let's build
something that
perceives.
Open to research collaborations, internships, and interesting problems in computer vision, machine learning & 3D vision.