Electronics and Communication Engineering undergraduate based in Bhubaneswar, India. I work on edge-AI and neuromorphic computing — bridging machine learning models with the hardware constraints they'll eventually run on.

Currently exploring Spiking Neural Networks (SNNs) and Hyperdimensional Computing (HDC) as low-power alternatives to conventional deep learning, with a focus on real-world, resource-constrained deployment (biomedical signals, edge vision).

• MarineSpike — converting a trained YOLOv8 detector into a spike-driven architecture for low-power inference, following recent work like SpikeYOLO (ECCV 2024) rather than naive post-hoc conversion.

• Neuromorphic ML: Spiking Neural Networks, Hyperdimensional Computing, ANN-to-SNN conversion
• Computer Vision: object detection pipelines, synthetic data generation, multi-object tracking
• Applied ML Engineering: iterative model debugging, dataset curation, honest evaluation practices

• NeuroHDC — Spiking-HDC framework for edge ECG classification; benchmarked against CNN/LSTM baselines across MIT-BIH, PTB, and STT datasets, with a documented data-leakage finding in standard CNN evaluation protocols.
• DeepOcean-AI — Underwater object detection and tracking (YOLOv8 + BoT-SORT), built through iterative real-world failure analysis rather than a single training run — documented false starts included.