Sunny Soni
I build machine-learning and AI systems, and I care about the part that usually gets left out: getting them to actually run somewhere real. On tiny edge hardware. Inside a privacy-first app. Against a game opponent that fights back.
Bengaluru, India. Open to internships and research.
I'm a B.Tech student in Digital Transformation at Atria University, Bengaluru. I build machine-learning and AI systems, and I like the part most projects skip: getting the thing to actually run somewhere real, whether that is a tiny edge board, a privacy-first app, or a game opponent that fights back.
My interests are agentic AI, robotics and embedded intelligence, and how you tell whether an applied ML system holds up outside its benchmark. I'm looking for research and ML internships, and building toward graduate study.
Four projects, each with something you can open.
A short idea of how each one works, the tech stack, the results, and the slides or code if you want to go further.
Featured work
Brain Tumor Segmentation on the Edge
Getting a multimodal-MRI tumour segmentation model to run on a 5-watt NVIDIA Jetson Nano, without losing the accuracy that makes it worth using.
Nyaya Sahayak
A privacy-first legal assistant for Indian practice. It keeps the language model on your own machine and still does the work end to end, built as an agentic RAG system.
More projects
Reinforcement Learning, Pong
Two DQN agents for Atari Pong, a plain baseline and a tuned version, built to show which change actually makes an agent learn.
Speech Restoration & Enhancement
A comparison of two classic noise-reduction methods, spectral subtraction and Wiener filtering, that starts from a failure and works out why one breaks where the other holds.
What I work with.
Machine Learning and AI
Generative AI
Computer Vision
Programming
Deployment and Tools
A few things worth mentioning.
Aspire Leaders Program, Finalist
Jun to Aug 2025One of 6,175 finalists from 36,050 applicants worldwide (top 17% of Cohort 3). A leadership curriculum designed by Harvard faculty, focused on AI and social impact.
Conference paper, ICNGSMCA-26
Jul 2026Presented the paper "Edge AI Framework for Brain Tumor Segmentation and Classification with Quantized Jetson Deployment" at the 2nd International Conference on Next-Gen Semiconductor Devices, Materials and Smart Computing Applications, Bengaluru.
Certificates