class AyushDas:
role = "AI & Machine Learning Engineer"
location = "India 🇮🇳"
focus = ["Multi-Agent Systems", "NLP Pipelines", "RAG", "LLM Applications"]
currently = "Building autonomous agents with LangGraph + LangChain"
learning = ["DeBERTa fine-tuning", "Agentic system design", "Vector DBs"]
open_to = "Collaborations on open-source AI/ML projects"
fun_fact = "I turn research papers into deployed products 🚀"|
5-agent LangGraph system — plans, searches, reads, self-critiques, and writes structured research reports. Dual-layer memory via ChromaDB. Live SSE streaming.
|
5-phase NLP pipeline for detecting contradictions in legal documents. DeBERTa-v3-large + FAISS alignment + cross-attention heatmaps for explainability.
|
|
Interactive 3D Pan-Cancer dashboard — TCGA RNA-Seq analysis identifying 19 lineage-specific transcription factors. Random Forest at 98.76% accuracy. Rediscovers known master regulators (HNF1B, GATA3, NKX2-1).
|
Production-grade carbon footprint platform — XGBoost ML predictions with conformal intervals, RAG sustainability advisor, agentic BoM decomposer. Real LCA data + explainable SHAP attributions.
|
Languages
AI / ML / NLP
Agentic & Vector
Web & Backend
I'm always open to interesting conversations about AI agents, NLP research, or collaborations.
