Aug 2022 – Mar 2023 · 7 mos

• Supervisor: Prof. Nitya Tiwari
• Worked on a novel approach that leverages signal processing and deep learning techniques to enhance the efficacy of self-supervised single-channel speech enhancement through noise-aware training of variational autoencoders.
• Integrated the speech-presence-probability-based Minima Controlled Recursive Averaging (MCRA2) algorithm to enhance model’s robustness in the presence of unseen noise conditions.
• Implemented the model architecture in PyTorch, achieving superior performance with a 5% improvement in aggregated metrics for NISQA and STOI compared to state-of-the-art models, under various noise scenarios.
• Nominated for the prestigious Best Undergraduate Thesis Award of the institute.

Jun 2022 – Sep 2022 · 3 mos

• Worked on the tokenization of venture capital by building a decentralized platform over the Polygon test network for startup founders to list their ventures and VCs to invest in them.
• Developed the token in Solidity, formulating tokenomics, platform design and overall financial architecture including fee structures, reward systems, and liquidity management strategies with the help of bonding curves.
• Built the frontend using React with features like crypto wallet authentication, portfolio dashboard, and venture investment. Integrated smart contracts and IPFS via Hardhat and Ethers.js.

May 2021 – Jul 2021 · 2 mos

• Developed an independent module using React, Relay and GraphQL to display customer enterprise locations on an interactive world map and integrated Google Places API for auto-complete search functionality.
• Implemented search optimization techniques like throttling, debouncing and lazy loading, reducing redundant API calls and improving DOM rendering efficiency by 20%,
• Collaborated closely with the UX team to turn wireframes and mockups into pixel-perfect designs. Enhanced the product’s responsiveness and accessibility, directly impacting user experience of over 200,000 global customers.