At Naresh i Technologies, I designed, trained, and deployed production AI models and agentic systems for NLP and automation across AWS and Azure. My RAG architectures used FAISS, ChromaDB, and Pinecone, achieving 95%+ retrieval accuracy and reducing hallucinations by 85%.
I fine-tuned GPT-4o, Claude, LLaMA, and Mistral models with LoRA/PEFT, and used MLflow for model versioning and output monitoring. I also developed FastAPI services with Docker and CI/CD pipelines, with streaming endpoints that achieved sub-2-second response times.
In my projects, I developed a document intelligence system for scanned, handwritten, and structured documents, and created multi-agent systems for workflow automation and research. I also worked as a Data Science Intern at Cognifyz Technology, preprocessing datasets and building classification models with Python, pandas, and scikit-learn.
