class RajeshKanade:
role = "Generative AI Engineer / Full Stack Developer"
company = "Studium Tech Pvt. Ltd., Pune"
education = "MCA @ Fergusson College (Autonomous) | CGPA: 9.10"
research = "Co-authored BERT-based Fake News Detection (NLP)"
looking_for = ["GenAI Engineer", "Full Stack / MERN Developer"]
location = "Pune, India 🇮🇳"
stack = {
"LLM Frameworks" : ["LangChain", "LangGraph", "LlamaIndex"],
"RAG & Search" : ["ChromaDB", "FAISS", "Pinecone", "BM25 Hybrid"],
"Evaluation" : ["DeepEval", "Ragas", "LangSmith", "LangFuse"],
"Backend" : ["FastAPI", "Node.js", "Python", "Docker"],
"Fine-Tuning" : ["LoRA", "QLoRA", "HF PEFT", "SFT"],
}
def current_focus(self):
return "Shipping production RAG pipelines & LangGraph agentic workflows 🔥"| Project | Description | Stack | Metrics |
|---|---|---|---|
| 🔍 PageMind — AI Web Assistant | Chrome extension with RAG pipeline for webpage Q&A & video transcript analysis | LangChain • LangGraph • ChromaDB • FastAPI • React | ⚡ Sub-2s latency · 📈 ~40% better relevance · 🧠 ~30% less hallucinations |
| 📄 Doc Intelligence Platform | Multi-modal document platform with hybrid search & agentic reasoning | LlamaIndex • AutoGen • ChromaDB • Redis • Docker | ⚡ ~45% faster queries · ✅ 0.87 faithfulness score |
| 🎓 Accreditation Insight Generator | LangGraph state machine for LLM-driven gap analysis and evidence mapping | LangGraph • LangChain • FastAPI • PostgreSQL • Pydantic | 📊 ~70% reporting efficiency gain · ⏱️ ~60% less manual work |
| 🎙️ AI Interview Coach | Personal voice-based interview coach — Whisper + VAD + FAISS RAG on resume | Whisper • FAISS • LangChain • GPT-4o-mini • React+Vite | ⚡ Sub-3s end-to-end response |
BERT-Based Fake News Detection Using NLP — Co-authored under Prof. Kalpana Joshi, Fergusson College (Autonomous), Pune. Applied BERT transformer for fake news classification achieving high accuracy on benchmark datasets. (2025)