AI engineering

Agents, MCP servers and RAG — in production, with guardrails

I build AI that works on real operational data: tool use into live systems, retrieval with evals, and the boring reliability work that keeps it safe.

Netsync

Agent

Production AI agent

Anthropic + OpenAI APIs with LangChain: multi-step reasoning, tool use, memory and guardrails for device, support and ops workflows.

Protocol

MCP

Tool server for LLMs

Structured query and action tools with secure calls into the database, telemetry and internal services.

Retrieval

RAG

Qdrant + evals

Training

SageMaker · Bedrock

01 / Ask

Ask my CV — live

Grounded in this site's case studies and CV, with sources.

Ask my CVAI answers from my real work, with sources

Beta

Ask me anything about Ahmed's projects, stack or availability.

What AI systems has Ahmed Mamdouh shipped to production?

A production AI agent on the Anthropic and OpenAI APIs (LangChain, tool use, memory, guardrails), an MCP server exposing internal systems to LLM clients, and RAG pipelines on AWS with Qdrant and eval harnesses — all at Netsync.

Does he do fine-tuning?

Yes — fine-tuning on S3, SageMaker and Bedrock, with eval harnesses to catch quality regressions.

Can he add AI to an existing product?

Yes. Typical work: an agent with tool calls into your APIs, RAG over your docs and data, guardrails, evals and cost tracking.

02 / Stack

The AI stack I use

Claude & Claude CodeAgents, skills, MCP

OpenAI APITool use, structured output

LangChainAgent orchestration

MCP serversSecure tools for LLMs

Qdrant · pgvectorRAG retrieval

SageMaker · BedrockFine-tuning, hosting

AI SDKStreaming UIs

SupabasePostgres, auth, vectors

03 / Stages

AI and real-time case studies

100K+concurrent devices

1-1 · 2023 — Now

ALTO device platform

Real-time device management for an IPTV platform: NestJS, Socket.io and MongoDB holding 100,000+ concurrent device connections.

NestJSSocket.ioMongoDBGraphQL

Read case study

MCPtools for LLMs

1-2 · 2024 — NowAI

ALTO AI agent + MCP

A production AI agent on the Anthropic and OpenAI APIs with tool use, memory and guardrails — plus an MCP server that gives LLMs safe access to internal systems.

LangChainAnthropic APIOpenAI APIMCP

Read case study

Add an AI agent to your product

Tool use, RAG on your data, guardrails and evals — scoped in one call.

Start a mission