You will get A full-stack AI SaaS MVP on Next.js + Supabase + TypeScript or FastAPI
Project details
I build AI-native MVPs as one coherent system — front-end, back-end, schema, and AI layer. You don't have to coordinate three different freelancers.
Two stack paths I'll recommend during discovery:
— TypeScript-only (default): Next.js handles front-end + back-end via API routes/Server Actions, Supabase for DB/auth, LangChain.js for AI. Cleaner ops, faster ship. This is the stack behind the 68,000-line PsHub monorepo.
— Next.js + FastAPI: when AI workloads need Python (custom embeddings, fine-tuning, LangGraph orchestration).
Recent shipped: PsHub — Next.js 16 + React 19 + Supabase monorepo on Cloud Run, with multi-agent AI built into the product, not bolted on. 64 tables, 350+ RLS policies for multi-tenant isolation from day one.
What sets this apart: most "full-stack AI" devs do one part well and copy-paste the rest. I architect the schema, API, front-end state, AND AI integration as a single coherent system — production AI products fail at integration boundaries, not at any single layer.
Two stack paths I'll recommend during discovery:
— TypeScript-only (default): Next.js handles front-end + back-end via API routes/Server Actions, Supabase for DB/auth, LangChain.js for AI. Cleaner ops, faster ship. This is the stack behind the 68,000-line PsHub monorepo.
— Next.js + FastAPI: when AI workloads need Python (custom embeddings, fine-tuning, LangGraph orchestration).
Recent shipped: PsHub — Next.js 16 + React 19 + Supabase monorepo on Cloud Run, with multi-agent AI built into the product, not bolted on. 64 tables, 350+ RLS policies for multi-tenant isolation from day one.
What sets this apart: most "full-stack AI" devs do one part well and copy-paste the rest. I architect the schema, API, front-end state, AND AI integration as a single coherent system — production AI products fail at integration boundaries, not at any single layer.
Programming Languages
HTML & CSS, Python, TypeScriptCoding Expertise
Cross Browser & Device Compatibility, Performance Optimization, DesignWhat's included
| Service Tiers |
Starter
$3,000
|
Standard
$6,500
|
Advanced
$10,000
|
|---|---|---|---|
| Delivery Time | 21 days | 42 days | 70 days |
Number of Revisions | 2 | 3 | 5 |
Number of Pages | 4 | 8 | 15 |
Design Customization | - | ||
Content Upload | |||
Responsive Design | |||
Source Code |
Optional add-ons
You can add these on the next page.
Additional Revision
+$300
Additional Page
(+ 3 Days)
+$300
Design Customization
(+ 7 Days)
+$800About Abdulrahman Assem Sameir
AI Engineer + Full-Stack | LangGraph, Voice AI, RAG, Next.js, FastAPI
Shibin al Kawm, Egypt - 4:10 pm local time
What I actually do:
— LangGraph multi-agent systems: Supervisor + Planner + worker agents with risk-classified human-in-the-loop gates. Tool-calling for 25+ tools per agent, streaming chat, and OpenRouter-based LLM gateway with model fallback so OpenAI rate limits don't break your product.
— Real-time voice AI: Pipecat + WebRTC + LiveKit pipelines with Deepgram, ElevenLabs, OpenAI, Groq, AWS. Multi-provider gateway with automatic failover so a Deepgram outage doesn't kill your call. Squad-based handoff across Twilio, Telnyx, and Daily for concurrent production calls.
— Hybrid RAG: BM25 + pgvector with cross-encoder reranking, semantic response caching on Redis, and auto-embedding pipelines. Shipped a 3-tier memory system for supplier-project matching. Vanilla vector search alone won't get you past 70% accuracy — I know what to build instead.
— Full-stack platform engineering: Next.js 16, React 19, Supabase, FastAPI, PostgreSQL with 350+ Row Level Security policies, pgmq async queues. The PsHub B2B marketplace I architected is a 68,000-line monorepo running on Cloud Run with Cloud Build CI/CD. I write the front-end, the back-end, the schema, and the AI layer — not just one slice.
— HIPAA-compliant clinical platforms: Firebase Genkit orchestrators, RAG over patient notes, encrypted audit trails, LiveKit voice agents for clinical staff. Built and shipped at AddisCare for U.S. healthcare.
— Web3 and automation: Crypto ecosystem at Profit 717 serving 50,000+ users — Python asyncio + aiogram + MongoDB + Redis backend, on-chain SHIBA payouts via smart contracts, multi-currency Web3 payments. Years of Playwright/Selenium scraping infrastructure work too.
Recent shipped work:
- Hubby — LangGraph v1 multi-agent platform on Next.js 16 / React 19 / Supabase (PsHub)
- Hams.AI — real-time voice AI backend (5 model providers, sub-400ms)
- AddisCare — HIPAA-compliant clinical decision support (USA)
- Profit 717 — crypto ecosystem serving 50,000+ users (Germany)
How I work: I write code, not slide decks. Working prototype within the first week so you see real progress fast. Daily 1-line updates over Slack/Discord. If something is going to slip, you hear about it the day I notice — not the week before launch.
Tech stack:
- Languages: Python, TypeScript, JavaScript, Rust, SQL
- AI/ML: LangChain, LangGraph, Pydantic-AI, Firebase Genkit, MCP, PyTorch
- Voice AI: Pipecat, LiveKit, WebRTC, Twilio, Telnyx, Deepgram, ElevenLabs
- RAG: pgvector, BM25, cross-encoder reranking, Redis caching
- Backend: FastAPI, Node.js, asyncio, SQLAlchemy, Pydantic 2.x
- Frontend: Next.js, React, React Server Components, Tailwind, Tauri
- Databases: PostgreSQL with RLS, Supabase, MongoDB, Redis, pgvector, pgmq
- Cloud: GCP, Cloud Run, AWS, Docker, Vercel, GitHub Actions
If you're scoping any of these — voice AI agent, multi-agent LLM system, RAG pipeline, HIPAA-compliant healthcare AI build, or a full-stack AI product MVP from front-end to back-end to model — that's exactly what I do every week. Drop the job details and I'll tell you what I'd build, in what order, and where the real risks are.
Steps for completing your project
After purchasing the project, send requirements so Abdulrahman Assem Sameir can start the project.
Delivery time starts when Abdulrahman Assem Sameir receives requirements from you.
Abdulrahman Assem Sameir works on your project following the steps below.
Revisions may occur after the delivery date.
Step 1
Discovery call (60-90 min) — confirm scope, MVP boundaries, AND stack choice (TS-only vs. mixed Python)
Step 2
Architecture document — system design, schema, stack rationale, deployment plan (you approve before build)