Aditya Shah | Full Stack & AI Engineer | React, Next.js, Python | Aditya Shah - Full Stack Developer

Associate Software Engineer · Contentstack

I turn language models into working software.

Full Stack & AI Engineer. I build the agents, tools and interfaces that sit between an LLM and the people who depend on it, from the model call to the pixel.

agent · reaction-to-ticket live

▸reaction_added :ticket: in #eng-bugs

▸reading thread · 14 messages

▸drafting ticket · claude

▸jira.create ENG-1432 "Login fails on Safari 17"

▸replying in thread with link

✓done in 2.1s

Work

Selected projects

All projects

Loading projects…

RAG Knowledge Assistant - AI by Aditya Shah

AI · 2026In progress

RAG Knowledge Assistant

Ask questions over your own documents and get answers grounded in them. Files are chunked and embedded into a vector index, and every answer cites the passages it came from.

AI Concierge - AI by Aditya Shah

AI · 2026In progress

AI Concierge

A conversational concierge for hotels and co-working spaces. Guests ask in plain language; the agent handles bookings, local recommendations and service requests, and hands off to staff when it should.

Lab

Agents, tools and experiments

Smaller things I'm building around language models. Some are shipped, some are half-finished on a branch.

  • Contentstack MCP Server

    An MCP server that exposes CMS entries, assets and workflows as tools, so Claude and Cursor can read, draft and publish content from inside the editor.

    TypeScript · MCP · Contentstack SDK

    Planned
  • Reaction → Ticket

    React to a Slack message with an emoji and an agent reads the thread, drafts a Jira ticket with Claude, files it and replies with the link.

    Next.js · Slack API · Jira · Claude

    Shipped
  • Agentic Workflow Builder

    Drag-and-drop nodes for API calls and LLM steps, wired together on a canvas. Trigger a run manually or from a webhook and watch each node execute.

    React Flow · Node.js · OpenAI

    Prototype
  • Agent Eval Harness

    Replay recorded agent trajectories against a rubric, grade them with an LLM judge, and track regressions per prompt version in a small dashboard.

    Python · Next.js · Claude

    Planned

Writing

Latest articles

All articles

Loading articles…