Uma Maheshwari Earati

LLM-Powered Semantic Image Search System

• Built an LLM-powered semantic image search system using multimodal models (CLIP) for text-to-image retrieval.
• Generated and indexed image & text embeddings (CLIP / BLIP / GIT) with FAISS and Qdrant for low-latency similarity search.
• Developed a RESTful backend with FastAPI and an interactive Streamlit front-end for search and visualisation.
• Enhanced search relevance using LLM-based query rewriting and optional metadata filtering (category, tags, colour).
• Containerised and deployed the application using Docker on Hugging Face Spaces / Streamlit Cloud.

Document Portal – AI-Powered Document Analysis & Chat System

• Designed and developed an AI-powered document portal using Python and FastAPI for document upload, analysis, and querying.
• Implemented document ingestion and preprocessing pipelines for both structured and unstructured data.
• Built an embedding-based document chat and retrieval system to enable accurate question answering over uploaded documents.
• Applied prompt engineering and reusable prompt templates to improve response quality and consistency.
• Containerised and prepared for AWS deployment using Docker, CI/CD with GitHub Actions, and production-ready practices (logging, exception handling, modular architecture).

LangGraph-Based Interview Preparation System (End-to-End GenAI Application)

• Designed and developed an end-to-end Generative AI system using LangGraph to orchestrate multi-step interview workflows.
• Implemented an agent-based architecture for automated question generation, web search, expert answer synthesis, and structured content writing.
• Integrated LLM tool calling, structured outputs, and persistent memory (checkpointing) for context-aware, multi-step responses.
• Built FastAPI back-end services to manage interview sessions, transcripts, and generated content.
• Containerised and deployed the application using Docker on AWS, with a strong focus on modular design, logging, error handling, and clean separation of core logic from production concerns.

• Joined the organisation as a Trainee and progressed into Team Lead–level responsibilities based on performance and commitment.
• Coordinated end-to-end product listing operations across multiple e-commerce marketplaces, ensuring accuracy and timely execution.
• Conducted website audits to identify content and listing issues and supported quality improvements.
• Managed email newsletter campaigns, including basic design, content preparation, and scheduling.
• Oversaw inventory coordination across marketplaces to maintain stock accuracy and availability.