Browse categories
Explore
Fiverr Pro
English
$
USD
I will add persistent memory to your ai agent with mcp and vector search
Giorgio G.
About this gig
Your AI agent forgets everything between sessions. I fix that.
I'm the author of rememori, an open-source agent-memory engine published on npm (~1,000 downloads in its first week), with a companion MCP (Model Context Protocol) server. I bring the same production-grade architecture to your agent.
What you get:
- Persistent semantic memory for your agent (ChatGPT, Claude, local Ollama or custom)
- MCP server setup so your AI tools remember across sessions
- Vector search with recency and importance ranking, tuned to your use case
- Clean TypeScript/Node.js code, tested and documented
How I work:
- Clear scope before starting, honest estimates
- Production patterns: clean code, tests on critical paths
- Async communication, regular updates
Tech: TypeScript, Node.js, OpenAI API, Anthropic Claude, Ollama (fully local and private setups welcome), vector search, RAG, MCP.
Not sure which package fits? Message me with a short description of your agent and I'll point you to the right one.
Get to know Giorgio G.
Giorgio G.
Backend Developer: NestJS, TypeScript, AWS and AI Integration
- FromItaly
- Member sinceJul 2026
- Avg. response time1 hour
Languages
Italian, English
Backend engineer with 4+ years building enterprise, multi-tenant SaaS platforms on AWS serverless. I design and ship production APIs in TypeScript, NestJS, AWS Lambda and Aurora PostgreSQL. Author of rememori, an open-source agent-memory engine on npm, including an MCP (Model Context Protocol) server. What I do well: clean, tested REST APIs; multi-tenant architecture; PostgreSQL schema design; AI/LLM integrations (OpenAI, Claude, Ollama, RAG, vector search); serverless AWS deployment. Clear communication, reliable delivery, honest estimates.
FAQ
Which AI models and platforms do you support?
OpenAI (ChatGPT/GPT API), Anthropic Claude, local models via Ollama, and any OpenAI-compatible endpoint. MCP clients like Claude Code and Cursor too.
Can everything run locally, without sending data to the cloud?
Yes. I regularly build fully local setups with Ollama — embeddings, storage and recall all on your machine. Nothing leaves your infrastructure.
Do I need an existing agent, or can you build the memory layer from scratch?
Both. I can plug memory into an existing agent/backend or design the memory architecture for something you're still building.
What do you need from me to start?
A short description of your agent (platform, language, what it should remember), access to the relevant repo or API keys for your setup, and your goal. I'll confirm scope before starting.
