I will build custom ai agents and multi agent systems


About this gig
Need an AI agent that can actually use your tools, data and workflows, not just answer questions?
I build and improve custom AI agents and multi-agent systems for real product and business workflows. I build around the full workflow, so the agent can retrieve the right context, use tools reliably, take structured actions and fit into the systems you already use.
I can help with:
- Support, knowledge, research and document agents
- RAG, memory and context-aware systems
- Tool calling, APIs and structured actions
- Multi-agent orchestration with specialist roles
- Agent evals, optimization and self-improvement loops
- Guardrails, reliability and failure handling
I work with TypeScript or Python using tools such as Mastra, LangGraph, CrewAI, OpenAI, Claude, OpenRouter and Hugging Face models.
Already have an agent? I can analyze failures and improve quality, reliability, latency or cost.
Message me before ordering with your goal, data and integrations. You do not need to know the right stack, I can help define the architecture and scope.
Get to know Aleksandar G
Full Stack Product Engineer
- FromGermany
- Member sinceAug 2026
- Avg. response time1 hour
Languages
English, German
My Portfolio
FAQ
What information do you need before starting?
I need a short description of what you want the AI agent to do, who will use it, what data it should work with, and any tools or APIs it needs to connect to. If you are unsure about the technical setup, that is completely fine — I can help define it.
Can you build an AI agent for my existing application?
Yes. I can integrate AI agents into existing web applications, SaaS products, internal tools, or APIs. I can work with your current architecture instead of rebuilding everything from scratch.
Can the AI agent use my own documents or business data?
Yes. Depending on the use case, I can connect the agent to documents, databases, APIs, vector databases, or other knowledge sources using RAG or similar retrieval approaches.
Can you build multi-agent systems?
Yes. I can build systems where specialized agents handle different responsibilities and work together through an orchestrated workflow.
Which AI models and frameworks can you work with?
I work with OpenAI, Claude, models available through OpenRouter, and Hugging Face models. I build with TypeScript or Python using tools such as Mastra, LangGraph, CrewAI, vector databases, and external APIs. The exact stack and model depend on the project.
Will I receive the source code?
Yes. You will receive the source code included in the agreed scope, together with instructions or documentation needed to understand and run the implementation.
Can the agent connect to external tools and APIs?
Yes. Agents can be connected to APIs, databases, search systems, business tools, custom backend functions, and other services so they can do more than simply generate text.
Should I contact you before placing an order?
Yes, especially for Standard, Premium, or custom projects. AI-agent projects can vary significantly depending on integrations, data, workflows, and complexity, so a short discussion helps make sure the scope is correct.
Can you improve an existing AI agent?
Yes. I can use traces, feedback and evals to find failure points and improve prompts, retrieval, memory, tools, routing, reliability, latency and cost.
Can you build self-improving AI agents?
Yes. I can build human-guided or bounded automated improvement loops that propose and test changes to prompts, retrieval, memory, tools or routing against defined metrics before accepting them.

