ML Engineer | LLMs · Generative AI · Agentic AI · RAG | Pre-Final Year | Seeking AI/ML Internships
I am a pre-final year Computer Science student at JSS Academy of Technical Education, Noida, specializing in Machine Learning, Large Language Models, Generative AI, and Agentic AI systems. I have hands-on experience building end-to-end AI applications — from Generative AI solutions and LangChain-powered RAG pipelines to LangGraph agentic workflows, fine-tuned transformer models, and voice-controlled AI agents. I work with modern agentic protocols including A2A and MCP, and deploy solutions using Streamlit with vector databases like FAISS and ChromaDB. Notable work includes: - A Multi-Source Agentic RAG System with autonomous retrieval strategy - A Voice AI Agent integrating speech-to-text with real-time LLM execution - A Fine-tuned RoBERTa model ranking Top 14% (76/560) on Kaggle Beyond technical work, I serve as Media Lead at the Entrepreneurship Development Cell (EDC) and have qualified the internal round of Smart India Hackathon. I am actively seeking AI/ML internship opportunities where I can contribute to impactful, production-grade AI systems. 📧 [email protected] 🔗 GitHub: https://github.com/Ishaan-Chaturved1
EDC JSSATE
7 mos
EDC JSSATE
1 yr 6 mos · Noida, Uttar Pradesh, India
B-Tech, Computer Science & Engineering
Expected 2027
Higher Secondary (12th)
2022
Score: 88.20%
Built a voice-controlled AI agent using Python, Streamlit, Groq, and AssemblyAI that converts speech into actions such as code generation, file creation, and summarization. Integrated speech-to-text, intent detection, response generation, multi-intent handling, session memory, and safe execution with user confirmation.
Built ML-based disease prediction using symptom-based NLP, implemented RAG with FAISS for medical knowledge retrieval, and added conversational memory for context-aware responses using Python, Scikit-learn, LangChain, FAISS, and LLM APIs.
Built a multi-agent RAG pipeline using LangGraph to orchestrate specialized agents for document retrieval, web search, and answer synthesis. Implemented autonomous retrieval strategy selection, FAISS and ChromaDB semantic search, OpenAI embeddings, LangChain RAG chains, streaming responses, and source citations through Streamlit.
Fine-tuned a RoBERTa transformer for binary classification of disaster-related tweets using HuggingFace Datasets and Trainer API. Achieved Rank 76/560 (Top 14%) on Kaggle with a score of 0.83573, improved from 0.79007, and evaluated using accuracy, precision, recall, and F1-score.
EDC JSSATE
Jan 2024 - Present · 2 yrs 4 mos
EDC (Entrepreneurship Development Cell)