YUNISH
UDAS
AI/ML Enthusiast · BSc. Computer Science & Information Technology
Building and experimenting with LLM applications, multi-agent workflows, and retrieval-augmented systems
SYS_PROFILE
ABOUT ME
I am a BSc CSIT student at Texas International College, Kathmandu, specializing in production-grade AI/ML systems. I move beyond simple demos to build reliable applications using Large Language Models (LLMs), multi-agent orchestration, and Advanced RAG.
With hands-on experience in LangGraph, LangChain, and CrewAI, I focus on building "agentic" systems equipped with guardrails, retry logic, and high retrieval accuracy. My portfolio includes a production-oriented Text-to-SQL assistant and a multi-agent content pipeline.
I am comfortable across the stack—from FastAPI backends to Streamlit interfaces—and I am currently seeking an internship to apply these skills to real-world, scalable AI challenges.
Agentic
- LangChain
- LangGraph
- CrewAI
- HuggingFace
- Ollama
ML / CV
- Pandas
- NumPy
- Scikit-Learn
- OpenCV
- Matplotlib
- Seaborn
Infra
- FastAPI
- Streamlit
- MySQL
- SQLite
- LlamaParse
- Git / GitHub
Languages
- Python
- JavaScript
- SQL
- C
- C++
PRODUCTION_WORK
Featured Projects
Text-to-SQL Assistant
Production NL-to-SQL agent with intent gatekeeping, destructive-query confirmation, and auto-retry logic (up to 2 attempts). Converts natural language to SQL, executes on SQLite, and returns grounded responses.
Stack: LangGraph · LangChain · SQLAlchemy · Groq · HuggingFace · Streamlit · SQLite
AI Blog Generator
Multi-agent content pipeline using a LangGraph router-planner with parallel workers. Generates structured technical blogs with optional live web research, Stable Diffusion image synthesis, and HITL export to Markdown/PDF.
Stack: LangGraph · LangChain · Groq · HuggingFace · Tavily · Stable Diffusion · Pypandoc
HR Policy Q&A System
RAG-based enterprise policy assistant for grounded document retrieval and question answering. Implements chunking, HuggingFace embeddings, ChromaDB indexing, and SelfQueryRetriever for metadata-aware retrieval.
Stack: LangChain · Gemini API · HuggingFaceEmbeddings · ChromaDB · Streamlit
data_object GitHub Profile
unixx0 open_in_newschool Education
BSc. Computer Science & Information Technology
Texas International College · Kathmandu
Operating Systems · Data Structures & Algorithms · Artificial Intelligence · Database Systems · OOP · Computer Networks · Discrete Mathematics · Probability & Statistics · Microprocessor & Assembly
+2 Science (NEB)
Global School of Science · Kathmandu