I'm an AI/LLM Engineer who builds systems that don't just respond β they reason, plan, and act.
My focus: Agentic AI β multi-agent systems using LangGraph and LangChain where autonomous agents collaborate, use tools, and solve real-world tasks end-to-end.
I just graduated in 2026 with a B.Tech in AI & Data Science (CGPA 8.37). Everything I know came from building things, breaking them, and shipping them.
riswan = {
"focus" : ["Agentic AI Systems", "LLM Fine-Tuning", "RAG Pipelines"],
"stack" : ["LangGraph", "LangChain", "LlamaIndex", "HuggingFace", "Groq"],
"deployment" : ["Docker", "AWS EC2/ECR", "FastAPI", "GitHub Actions"],
"currently" : "Building multi-agent systems that work while I sleep π€",
"fun_fact" : "I love Docker more than most people love coffee β"
}| System | What It Does | Stack |
|---|---|---|
| π₯ Agentic Medical AI | Multi-agent RAG over medical datasets with context-aware reasoning | LangGraph Β· Pinecone Β· Groq Β· Docker Β· AWS |
| π Autonomous Research Agent | Researcher + Analyst + Writer agents that produce full reports | LangGraph Β· Tavily Β· Groq Β· Streamlit |
| π MLOps Vision Pipeline | End-to-end YOLOv7 pipeline: ingest β train β evaluate β deploy | MongoDB Β· AWS S3/EC2 Β· MLflow Β· GitHub Actions |
| π€ LLM Fine-Tuning | Domain-adapted LLMs using LoRA & QLoRA | Unsloth Β· HuggingFace Β· Weights & Biases |
Core AI & LLM
Agentic & LLM Orchestration
Vector Databases & Retrieval
MLOps & Deployment
Data & Infra
User Input
β
βΌ
βββββββββββββββββββββββββββββββββββββββ
β Supervisor Agent β β LangGraph orchestrator
β (Plans tasks, routes to agents) β
ββββββββββββββββββ¬βββββββββββββββββββββ
β
ββββββββββββββΌββββββββββββββ
βΌ βΌ βΌ
ββββββββββ ββββββββββ ββββββββββββ
βResearchβ βAnalyst β β Writer β
β Agent β β Agent β β Agent β
β(Tools) β β(Python)β β(LLM out) β
ββββββββββ ββββββββββ ββββββββββββ
β β β
ββββββββββββββ΄ββββββββββββββ
β
βΌ
Final Output β
(Report / API Response / Action)
- π Cyclic LangGraph agents β agents that retry, self-correct, and loop until success
- π§ Long-term agent memory β persistent state across sessions
- βοΈ LLM fine-tuning on domain data β LoRA + QLoRA with Unsloth
- π Deploying agentic systems as APIs β FastAPI + Docker + AWS



