AWS_LLM
I completed Builders Skill Sprint and it was good learning for me. I am beginner in AI agents so I didnt understand many things at first. I build simple agent using Strands SDK and Ollama. Then I added tools like calculator, weather and age calculator. After that I used memory with Mem0 and FAISS so it can remember things. MCP part was little hard for me but I tried and it worked. I learned basics of AI agents step by step.
I recently finished Builders Skill Sprint and it was really good experience for me. I am still beginner in AI agents so many things I dont understand in starting but later slowly it make sense little bit.
In Challenge 1 I build simple agent using Strands SDK and Ollama llama3.2:3b. I was confused like how model is working and how python code is talking with AI model. I just run code many times and try to change small things and then I understand little.
Then Challenge 2 I add tools like calculator tool, weather tool and age calculator tool. This part was more interesting because agent is not only chatting but also doing calculation and answering weather etc. I got some errors like tool not found and import issue but I fixed by checking again and again.
Challenge 3 was memory thing using Mem0 and FAISS. This was little hard for me. But when agent remember my name like I said “my name is Thamarai” and later it answer correct I was like wow it is working. Sometimes it forget also so I re-run and check code.
Challenge 4 I combine everything like tools + memory + chat loop. This was messy for me because many things together and sometimes terminal error coming. But I learned how full agent works in real structure.
Challenge 5 MCP server thing was most hardest for me. First time server not starting properly and I dont know why. I check youtube and docs and again try and then it worked. MCP is like connecting external tools to agent which is powerful but little confusing for me at start.
Overall what I learn from this sprint is:
- AI agents is not just chat bot, it is more than that
- tools make agent useful in real life
- memory is important so agent can remember user things
- MCP is advanced thing but powerful
- learning by doing is best method because I make many mistakes and then understand
I still not expert and I feel I need more practice but this sprint help me a lot to understand basics of AI agents and how real systems build.
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article