Shorya Sharma

Designing and developing enterprise-grade Generative AI applications and LLM-powered solutions for large-scale client deployments. Building Agentic AI workflows with multi-agent orchestration, tool use, Azure Agent SDK, and MCP (Model Context Protocol) integrations. Architecting RAG & Advanced RAG pipelines on Azure OpenAI + Azure AI Search, plus LangChain / LangGraph with vector DBs (Pinecone, Weaviate, FAISS, Chroma). Implementing Responsible AI & safety: Azure AI Content Safety, content filtering, guardrails, PII detection, prompt-injection defense, and bias mitigation across GenAI products. Building LLM evaluation pipelines with RAGAs, Promptfoo, and LLM-as-Judge frameworks for production quality assurance and regression testing. Setting up end-to-end observability & monitoring for LLM apps with LangSmith, Langfuse, Prometheus, and Grafana — tracking token usage, latency, cost, and quality metrics. Driving prompt engineering, LLM fine-tuning (LoRA / QLoRA), and evaluation frameworks for domain-specific GenAI use cases. Engineering production-ready GenAI services with FastAPI, Docker, CI/CD, and dual-cloud deployment on Azure (Azure ML, Functions, App Service) and AWS. Building data engineering pipelines with Azure Data Factory, Synapse, and Databricks to power GenAI & RAG use cases at scale. Collaborating with cross-functional teams to take GenAI products from POC to production and operationalization.

Built next-gen Agentic AI applications for identity & biometric solutions using LLMs and multi-agent systems. Implemented MCP (Model Context Protocol) and Agentic AI frameworks for tool-augmented, context-aware multi-agent workflows. Designed RAG pipelines with LangChain, LangGraph, and vector DBs (FAISS, Weaviate, Pinecone, Chroma) for real-time knowledge retrieval. Fine-tuned domain-specific LLMs with HuggingFace Transformers, TensorFlow, and PyTorch. Engineered production-grade APIs & AI services with FastAPI, Docker, and CI/CD pipelines. Drove multi-modal AI integration across face, fingerprint, iris, and voice biometric modalities.

Developed and fine-tuned cutting-edge GenAI models (Mistral, LLaMA, GPT, Gemini) for high-performance NLP solutions. Engineered Agentic AI systems with multi-agent collaboration for autonomous decision-making and workflow optimization. Drove POCs & MVPs from prototype to production deployment and operationalization. Built ML & DL models with scikit-learn, Keras, TensorFlow for face recognition, fire detection, crowd / congestion detection, and surveillance. Integrated RAG pipelines into GenAI applications for contextually accurate generation. Deployed scalable AI solutions with FastAPI & Docker, ensuring smooth backend integration.

Internship Certificate (9) (1).pdf