Hamza Nafasat
AI Developer | RAG Chatbots, AI Agents & Next.js SaaS
New to Contra
Hamza is ready for their next project!
Echo is a multi-tenant SaaS for AI customer support with realtime chat, an AI voice agent, and AI automation. I was the primary developer on the full build, frontend, backend, and the AI layer. The AI is the core. It runs OpenAI, Claude, Gemini, and Grok through one multi-model setup, so a client can switch providers without a rewrite. A RAG pipeline connected to a vector database grounds every answer in the client's own content, so the chatbot never returns generic output. VAPI powers the voice agent, so customers can speak to support on a live call. Each tenant gets its own AI agent built on its own documents. The stack is Next.js 15 and React 19 inside a Turborepo monorepo, with separate apps for the dashboard, the embeddable chat widget, and backend services. Realtime chat runs on Convex. Clerk handles auth. API keys are encrypted per tenant through AWS Secrets Manager, so no two clients share credentials or data. Launch day held 60 live conversations at once with zero dropped sessions. What the client gets: an AI chatbot and voice agent that answer from their own content, work across multiple LLM providers, and stay isolated and secure per tenant.:
1
96
I designed and built a Sentry-style error tracking SaaS from scratch as the Next.js developer and AI integration developer behind it, to prove out multi-tenant architecture and AI integration. Admins manage sub-users with per-project access control. Each admin attaches their own OpenAI, Claude, or Gemini key, encrypted with AES-256-GCM, and picks a model per project for AI fix suggestions. I built two SDKs (Node.js and browser, with a React error boundary) plus a CLI that uploads source maps on build. Errors resolve to original source code, not minified output, the same technique Sentry uses.
1
106
Meet AI is a video conferencing SaaS where AI agents join live meetings as active participants and respond in real time. I built the full application stack.
The hard part is the real-time layer. The backend provisions an AI agent the moment a meeting starts and keeps it running for the full call. Speech runs through a transcript pipeline and generates responses through the OpenAI API in near real time, so the agent replies while the conversation is still moving, not minutes later. After the meeting, the backend writes a structured summary through the same API, handled async by Inngest so nothing blocks the live app.
The stack is Next.js 15 and React 19 with the Stream Video and Stream Chat SDKs, tRPC and Drizzle ORM on Neon PostgreSQL, and Better Auth across the stack.
The platform runs 30 simultaneous live sessions, each with its own active AI agent, with no backend slowdown.
What the client gets: live AI agents that take part in real meetings, plus automatic summaries, on a stack built to hold many sessions at once.
2
75
Nodebase is a workflow automation SaaS where users build AI workflows visually with no code, backed by a custom execution engine. I built the full product end to end, alone.
The visual builder runs on React Flow inside Next.js 15. Users connect trigger nodes, AI nodes, API nodes, and conditional logic into full pipelines. The backend engine uses topological sort to resolve node dependencies before running each step, so nothing fires out of order. Inngest handles background jobs, retries, and scheduled triggers. Webhooks let outside services start a workflow by hitting an endpoint.
This is AI automation in practice. OpenAI, Claude, and Gemini run as built-in AI nodes next to Slack, Discord, and Stripe. A user can route data through an LLM, act on the result, and pass it to the next step, all without code. Credentials are encrypted and pulled at runtime, so keys never sit in plain text.
Prisma with Neon PostgreSQL handles workflow ownership, execution history, and job logs. tRPC connects frontend and backend with full type safety.
What the client gets: a no-code platform where non-technical staff build their own AI workflows, on an engine reliable enough to run them on a schedule without supervision.
1
88
Form Flow is a B2B SaaS for enterprise and financial teams where AI handles the full form logic. I built the AI layer that turns plain English into working software behavior.
The rule engine sends an admin's plain-English input to the OpenAI API, which generates executable JavaScript handlers from the description. Those handlers run inside a vm2 sandbox on Node.js, so AI-generated code executes without touching the rest of the app. It works like a locked box: code runs inside, nothing escapes. Rules chain in order and trigger status changes, alerts, and display actions. Getting output consistent enough to run in production without malformed handlers took real iteration, the careful side of AI integration.
A RAG layer reads the surrounding form structure and the exact field a user is on, then sends a prompt scoped to that field, so the AI assistant returns an answer that fits the question instead of something generic. On financial intake forms, this cut abandonment with no human agent involved.
Branding automation pulls colors, logos, and typography from any URL for white-label deployment. Identity verification runs through IDmission with live face matching for compliance.
What the client gets: non-technical admins define complex form logic in plain English, an AI assistant guides users field by field, and AI-generated code runs safely inside a sandbox.
2
78
RouteMind is a multi-tenant fleet management SaaS with live GPS tracking and an AI voice agent for operators. I built the full stack and the AI layer end to end.
The hardest part was the data architecture. Structured records like fleet profiles, driver data, and route history sit in MySQL. The high-frequency GPS event stream, which would choke a relational setup, runs through MongoDB instead. Vehicle positions render on a live map with latency low enough that movement feels real-time. That split is what kept the dashboard fast while thousands of GPS events landed every minute.
The AI layer is a voice agent built on a RAG pipeline and the OpenAI API. Operators call in, and the agent retrieves from a vector database of indexed fleet policies and route procedures before answering, never generic output. Questions outside its scope trigger a clean handoff to a human.
The platform is fully multi-tenant with strict data isolation between businesses. Roles and permissions are set from the dashboard, so onboarding a new operator needs no code change.
What the client gets: live fleet visibility, an AI agent that answers policy questions without tying up staff, and an architecture that holds up as the fleet grows.
1
4