AI / ML research · 2025
Agentic LLMs with Reflection & Tool Integration
Agentic reasoning in Llama 3 with iterative self-reflection and external tool use via the Groq API.
- Llama 3
- Groq API
- ReAct
- HTTP Tools
The problem
Make LLM outputs more reliable through self-reflection and tool use.
My role
Researcher & engineer (MS project).
What I built
Implemented agentic reasoning patterns in Llama 3 with iterative self-reflection and external tool interaction, built on the Groq API with HTTP tool integration and IPython.
Architecture & stack
- Llama 3 via the Groq API
- Iterative self-reflection loop
- HTTP tool integration; IPython workflows