Manshverse
An AI chat product for web and Android. Prompt intent routes a request to a suitable model; personas shape the conversation.
I’m Sparsh, a Computer Science student and developer from Bengaluru. I build web and mobile products, following ideas from the interface into the systems underneath.
See selected workI’m from Bengaluru and studying Computer Science. I build software across web and mobile, and I’m curious about how information moves through a product, how it behaves when a connection disappears, and how small interface decisions can make a tool easier to trust.
That curiosity has led me from AI chat and word discovery to conversation archives and offline-first expedition tools. I work across product thinking, interface implementation, APIs, and data—then keep refining until the pieces feel coherent.
A few products, from the first interaction to the details that make them work.
An AI chat product for web and Android. Prompt intent routes a request to a suitable model; personas shape the conversation.
A private visual way to explore WhatsApp and Discord history through activity, participants, and milestones. Analysis runs in the browser; cloud storage is optional.
EARLY DEVELOPMENT · USE SYNTHETIC CHAT DATA
An offline-first expedition and supply operations prototype for polar research stations. I led a six-person team and worked on the interface, local data, and planning.
TEAM PROTOTYPE · STATION DATA IS SIMULATED
A language experiment for feelings and ideas that are difficult to name. Describe one; Lexivoid asks Mistral for a word to explore.
A personal archive for writing down a moment and returning to it later.
A student developer community where I help lead local events and maintain the Bengaluru chapter site.
A React Native chat app with a Node.js and Express backend, Socket.io live updates, and Prisma data access.
Led a six-person Dev React team building Polaris, an offline-first prototype for polar expedition planning.
Polaris case study ↗Leading the Bengaluru fork of a youth-led builder network, running offline hackathons, build nights, and sprint sessions across the city.
Community site ↗Built a multi-model AI ecosystem with a semantic router that selects an LLM based on prompt intent; led the product and full-stack build.
Manshverse case study ↗Implemented deep learning architectures from scratch in PyTorch, progressing from RNNs to full Transformer models.
Research notes ↗Building web and mobile interfaces with React, Next.js, React Native, Vite, and HTML/CSS/JavaScript.
Connecting client applications to server routes, WebSockets, databases, and hosted services with Node.js, Express, and Firebase.
Working with model APIs from Groq, Gemini, and Mistral, and studying transformer fundamentals with PyTorch.
A few details are enough to start. I’ll reply by email to discuss the next step.
Prefer email? sparshsharmadev@gmail.com · Track an inquiry