Simran C- Senior Data Engineer

Major Projects

Intelligent Document Classification & Extraction Pipeline

Led the end‑to‑end development of an AI‑powered workflow for automatic document segmentation, classification, and data extraction. Implemented computer‑vision PDF splitting, integrated OCR with LLM‑driven field extraction, and added rigorous field‑level validation. Deployed the modular, scalable solution on Google Cloud Platform to support enterprise performance and easy future extensions.

Spain

Frontend:

FastAPI

Backend:

LangChain
Python

AI/ML/NLP:

Computer Vision
OCR
OpenAI GPT

Cloud Infrastructure:

GCP

AI Task Execution System with Agno Framework

Designed and developed a modular AI-powered task execution system using the Agno framework to automate daily productivity tasks via natural language commands. Architected multi-tool orchestration to enable seamless execution of actions like email handling, calendar management, shell operations, and data retrieval. Integrated specialized tools with a user-centric design, ensuring intuitive mapping from user input to backend operations. Rigorously tested for reliability, speed, and adaptability across diverse task scenarios.

USA

Frontend:

Natural language interface

Backend:

Agno framework

AI/ML/NLP:

multimodal workflow coordination
NLP

Integrations:

CalendarTools
EmailTools
MapTools
NewspaperTools
ShellTools
WikipediaTools
YFinanceTools
YouTubeTools

Design Anything – Advanced Image Editing System

Developed an AI-powered image editing platform enabling seamless content replacement within images. Designed the integration architecture combining Segment Anything Model (SAM) for precise segmentation with Stable Diffusion’s inpainting for high-quality visual edits. Focused on performance tuning, usability enhancements, and iterative improvements based on user feedback to deliver a robust and intuitive editing experience.

India

Backend:

Python

AI/ML:

Segment Anything Model (SAM)
Stable Diffusion

Image Processing:

OpenCV

Intelligent Context Manager for LLMs

Designed and built a Model Context Protocol (MCP) to dynamically manage and retrieve contextual memory for large language model applications. Implemented FAISS-based vector retrieval using embeddings, schema-driven validation with JSON Schemas, and modular pipelines for context ingestion, merging, and token overflow handling. Developed versioning and scalability features to support enterprise-level integration and personalized AI agent interactions.

Germany