Nov 2025 – Present · 6 mos

Jul 2025 – Aug 2025 · 1 mo · Seoul, South Korea · Remote

• As part of Hopae’s hConnect credential-orchestration initiative, move national eID providers from desk research to integration-ready status.
• Convert policy and provider documentation into concrete integration criteria, readiness checklists, and risk/feasibility notes; coordinate with issuers/regulators and delivery partners.
• Stand up sandbox POCs and define baseline security, observability, and data-handling requirements and acceptance tests for handoff to the core engineering team.
• Author runbooks and test plans, set go/no-go gates, and establish traceable audit logs to ensure repeatable, auditable production integrations.

Feb 2025 – Apr 2025 · 2 mos · Bengaluru, Karnataka, India · Hybrid

• Project: DevTrack – Developmental Assessment Platform with Smart Scoring & Active Learning
• Designed and built a FastAPI backend that powers a comprehensive assessment API, handling question input, keyword management, response scoring, caregiver analytics, and milestone tracking via endpoints like /question, /score-response, and more .
• Delivered a multi-tier scoring system: first-pass LLM-based scoring (when available), fallback keyword logic (with negation detection and milestone-specific patterns), and ensemble scoring to resolve uncertainty or edge cases .
• Architected an active learning pipeline that surfaces top‑priority examples for expert review—prioritizing by uncertainty, disagreement, novelty, and coverage—to continuously refine the model .
• Integrated Reason-to-Retrieve (R2R) for enhanced borderline case handling with context-aware retrieval-based reasoning, yielding ~40% accuracy improvement on edge cases .
• Created an interactive reactive web demo (React/HTML) for stakeholders to explore predictions, feedback flows, and active learning outcomes .
• Built a robust testing and reporting framework: end-to-end pytest suite, performance dashboards, and test-run logs delivered .
• Delivered production-grade documentation and diagrams, including quick-start guides, README, API docs, testing docs, and architectural PDFs .
• Impact: Built a robust, adaptive system to analyze caregiver-child developmental assessments—accurate, transparent, and capable of self-improvement through expert-in-the-loop learning. Perfected both scoring logic and workflow for continuous model validation.

Oct 2024 – Dec 2024 · 2 mos · Bengaluru, Karnataka, India · Hybrid

• Project: Sports Analyser using Deep Learning – Real-time Football Match Analytics System
• Developed a Python-based football analysis system capable of detecting, tracking, and annotating players, referees, and the ball from match footage in real time.
• Tech Stack: YOLOv5, OpenCV, Python, NumPy, Pandas, Matplotlib, SORT tracking algorithm, HPCC Systems for big data integration.
• Designed and implemented an object detection module using YOLOv5, fine-tuned with football-specific datasets for improved accuracy.
• Built a tracking module (SORT) to handle occlusions and fast-moving objects, ensuring reliable player and ball tracking.
• Developed video annotation functions with color-coded overlays, bounding boxes, IDs, and event markers for intuitive visualization.
• Integrated heatmap generation to visualize player movements, high-traffic zones, and tactical formations.
• Calculated performance metrics including player speed, distance covered, possession time, and ball trajectory.
• Optimized the system for real-time processing with GPU acceleration and fine-tuned detection thresholds for latency reduction.
• Key Outcomes:
• Delivered a fully functional, real-time football analytics MVP adaptable for both live match analysis and post-match review.
• Strengthened skills in computer vision, sports analytics, model training, and optimization.
• Demonstrated potential for cross-sport adaptability (basketball, cricket, hockey) and future enhancements like predictive analytics and AR integration.

Nov 2023 – Nov 2024 · 1 yr · Bengaluru, Karnataka, India · On-site

Jul 2023 – Nov 2023 · 4 mos · Bengaluru, Karnataka, India · Hybrid

• Project: Decentralized file storage system with NFT authentication.
• Developed smart contracts for file storage and NFT minting.
• Deployed on several blockchains, including Ethereum, Polygon, Solana, and Hyperledger Besu.
• Developed a private QBFT chain for enterprise-grade file storage in Hyperledger Besu.

Jan 2023 – Jun 2023 · 5 mos

Sep 2022 – Jun 2023 · 9 mos

Nov 2022 – Dec 2022 · 1 mo · Bengaluru, Karnataka, India · Hybrid

• Project: Decentralized Spotify – NFT-powered Music Streaming DApp on Polygon
• Completed during the “Buidl for Web3” internship with Coding Club, RVCE & BOBDAO (Nov 2022 – Dec 2022).
• Built a full-stack Web3 music streaming platform with NFT integration, empowering artists to mint music NFTs and stream content directly to users without intermediaries.
• Tech Stack: Solidity (smart contracts), React.js, Node.js, CSS, Moralis SDK, Alchemy SDK; deployed on Polygon Mumbai Testnet.
• Developed smart contracts to mint NFTs from MP3 files, ensuring unique ownership, secure metadata storage, and verifiable royalty payouts for creators.
• Designed a Spotify-inspired UI for seamless audio streaming, podcast hosting, and integrated NFT purchasing.
• Integrated wallet connectivity and blockchain-based payment flows, leveraging decentralized infrastructure for security and transparency.
• Researched and addressed challenges in tokenized royalty distribution and blockchain-payment integration for real-world scalability.
• Key Outcomes:
• Delivered a working MVP demonstrating NFT ownership, royalty potential, and a user-friendly decentralized streaming experience.
• Strengthened expertise in blockchain architecture, Solidity, dApp deployment, and Web3 integration.
• Gained practical experience with Polygon, Web3 APIs, decentralized storage, and on-chain payment models.