This is frozen ninja

Darshan Wagh  • Peerlist

-Built an AI-powered PDF reader with Adobe’s PDF Embed API. -Added context-aware recommendations (<2s, >80% accuracy). -Created “Insights Bulb” (LLM-powered) for key takeaways & contradictions. -Developed Podcast Mode: 2–5 min narrated summaries. -Delivered Dockerized prototype with offline support.

Adobe Dot | Finalist at Adobe India Hackathon

-Developed a full-stack web application using React, Express.js, and PostgreSQL that functions as a digital Pokédex. -Implemented interactive browsing, advanced search features, and detection capabilities to deliver an engaging user experience for Pokémon enthusiasts.

Pokedex | Winner of PICT TechRush'24

-Built an AI-powered web platform using React.js and TypeScript that validates and highlights errors in termsheets to ensure security compliance. -Deployed a DeepSeek model on Bedrock, orchestrated with Kubernetes, and integrated SQS and Lambda functions to support scalable, fault-tolerant, and parallel processing for high-performance validation

Sure Sheets | Finalist at Barclays Hack-o-Hir

-Built an AI-driven evaluation system in Python that compares outputs from multiple large language models to identify bias and measure response quality. -Leveraged sentence embeddings and NLTK within Google Colab to provide structured, data-driven assessments of fairness and accuracy in model responses.

UnbiasedGPT | Winner at PCCOE MLSCOE'24

-Created a real-time stock market simulation platform with 1000+ participants that allows participants to trade virtual stocks and compete based on performance. -Implemented live data streaming, live news delivery, portfolio analytics, and competitive leaderboards to deliver an engaging and interactive trading experience.