s
sdass1918

Sudipta Das

@sdass1918

AI Full Stack Developer

Inde
Anglais, Hindi
Certaines informations sont présentées en anglais.
À propos de moi
Hi, I'm Sudipta, an AI Full Stack Developer specializing in AI-powered web applications, LLM integrations, and scalable backend systems. I'm a Google Summer of Code 2026 contributor at the Internet Archive, building AI features using Gemini APIs. I build SaaS MVPs, AI agents, Chrome extensions, and full-stack applications with React, Next.js, Node.js, TypeScript, Redis, Docker, and PostgreSQL. I focus on clean, scalable, production-ready solutions and clear communication. Let's build something amazing! 🚀... Plus d’infos

Compétences

s
sdass1918
Sudipta Das
hors ligne • 

Voir mes services

Sites web IA & Logiciel
I will build ai saas web apps using react nextjs nodejs fastapi and llms
De50 $US / projetPlus de détails

Portfolio

Expérience professionnelle

Internet_Archive

Google Summer of Code Contributor

Internet Archive • Temps partiel

Apr 2026 - Present • 3 mos

Selected as a Google Summer of Code 2026 Contributor at the Internet Archive to build AI-powered features for the Wayback Machine, one of the world's largest digital archives. My project focuses on integrating Chrome's built-in on-device Gemini AI APIs to make archived web pages more useful and accessible. During the program, I engineered production-ready features including AI-powered page summarisation, page quality analysis, multilingual translation, streaming AI responses, multimodal screenshot analysis, structured JSON outputs, and interactive archive dashboards with intelligent caching. Working on this project required much more than implementing features. Since Chrome's AI APIs were still evolving, I frequently had to work directly from documentation, prototype new approaches, and debug behaviours that had very little existing guidance. One of the most challenging problems involved redesigning the AI session architecture after discovering that shared model sessions leaked conversational context across independent features. Solving it required carefully studying the API's behaviour, redesigning the inference pipeline using isolated session cloning, and improving both correctness and responsiveness. Throughout the program, I collaborated closely with experienced open-source mentors through weekly design reviews, discussed architectural trade-offs, incorporated code review feedback, and continuously refined my implementations before submitting pull requests. This experience strengthened my ability to design production-ready systems, debug complex distributed and AI-driven applications, and write maintainable code for a real-world open-source project used by millions of people worldwide.