Browse categories
Explore
Fiverr Pro
English
$
USD
I will build production ready autonomous ai agents and multi agent systems
Murtaza
About this gig
Tired of AI that breaks on the first edge case? I build autonomous agents as engineered systems, not one-shot prompts.
Every agent runs a harness loop: Recall -> Think -> Decide -> Act -> Observe -> Repeat. It pulls memory, plans the goal, decides if a tool is needed, calls it, reads the result, and re-plans until done. For long runs I add context compaction, keeping only what the next step needs.
WHAT I BUILD
- LangGraph stateful agents with memory and conditional logic
- CrewAI multi-agent systems (planner, executor, verifier)
- Agents that use real tools: search, APIs, databases, files
- Self-correcting pipelines that retry on failure
- LangSmith tracing and evaluation built in
- Agentic means goal + planner + tools + state + observation + adaptation, inside guardrails.
STACK
LangGraph, CrewAI,Autogen, LangChain, FastAPI, Python, GPT and Claude, LangSmith,Redis,MCP
USE CASES
- Research and report agents
- Data processing automation
- Assistants with real tool access
- Multi-step business workflows
Good fits get clean, modular code, error handling and docs. Message me first and I will scope your workflow free.
Get to know Murtaza
Murtaza
AI engineer for autonomous agents and RAG systems
- FromPakistan
- Member sinceJun 2023
- Avg. response time2 hours
Languages
Urdu, English
I build AI agents and RAG systems that hold up in production. Most AI is a one-shot prompt hoping for the best. I engineer harnesses: recall, think, decide, act, observe, repeat until the goal is met. For RAG I retrieve, rerank, gate, then answer or honestly abstain. Python, FastAPI, LangChain, LangGraph, Autogen, CrewAI, LangSmith tracing, vector DBs (pgvector, Pinecone, Qdrant, Supabase). I ship evaluated, cost-aware systems, not demos. Message me and I will scope your project free.
My Portfolio
FAQ
What is an agentic AI workflow?
An autonomous system where the agent reasons, decides, uses tools and executes multi-step tasks instead of just replying to prompts
Will long tasks stay accurate?
Yes. I apply context compaction so the agent keeps signal high and cost low across long runs.
Is this production ready?
Standard and Premium are built for production, with error handling, retries, logging and scalable structure
Which AI models do you use?
GPT, Claude, Gemini or open-source models, chosen for your use case and budget
Do I get the source code?
Yes, full source code with clean structure and documentation.
LangGraph or CrewAI?
LangGraph for stateful single or multi-agent logic, CrewAI for role-based teams. I recommend the best fit
Do you add LangSmith tracing?
Yes. Standard and Premium include LangSmith tracing and evaluation so every run is observable and debuggable.
Can the agent use my documents?
Yes. I build RAG-enabled agents grounded in your documents, databases or knowledge base
Should I message before ordering?
Yes. A short scoping message ensures the right package and a smooth delivery.
What do you need to start?
Your workflow description, access to the APIs or tools the agent should use, and the data sources it should read
