Sanjay S.

AI Developer and AI Voice Agent Developer | LLM fine-tuning, RAG, AI agents, Amazon Bedrock | Python developer since 2017 | MCA in Computer Science I help startups and SaaS teams turn large language models into working products: voice agents that talk to customers, AI chatbots that know your business, and AI agents that take real work off your team. WHAT I BUILD • AI voice agents: real-time voice bots using Deepgram, Amazon Polly and the OpenAI API for customer support, outbound calls and appointment booking • AI agents: multi-agent systems with CrewAI, AutoGen and LangChain, including tool calling and workflow automation • RAG and AI chatbots: retrieval augmented generation over your documents with LangChain and vector databases, so answers come from your own data • LLM fine-tuning: LoRA, QLoRA and PEFT on Llama 3 and Mistral for domain tasks such as Q&A, persona writing, legal and medical text • LLM deployment: Amazon Bedrock, Azure OpenAI, and self-hosted open-source models with vLLM and TGI on AWS, GCP or RunPod • Generative AI apps: Python back ends (FastAPI, Django, Flask) with React front ends, plus Streamlit and Gradio demos SELECTED AI WORK • MindReader AI: a personality-based sales coaching chatbot • KeepersVolt: an AI-powered application • AI product image generation and a virtual try-on shopping prototype TECH STACK LLMs: OpenAI API, Azure OpenAI, Amazon Bedrock, Llama 3, Mistral, Mixtral AI frameworks: LangChain, CrewAI, AutoGen, Hugging Face Voice AI: Deepgram, Amazon Polly Back end: Python, FastAPI, Django, Flask, Node.js, PostgreSQL, MongoDB, Firebase Cloud and DevOps: AWS, GCP, Azure, Docker, Kubernetes, CI/CD HOW I WORK • A short scoping call and a written plan before I start • Small, testable milestones with a demo at each step • Clean, documented code and a handover you can run yourself Tell me about your use case and I'll suggest the simplest AI solution that will actually work for it.