Skip to content
View ParasSondhi's full-sized avatar

Achievements

Achievement: Pull Shark

Achievements

Achievement: Pull Shark

Block or report ParasSondhi

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
ParasSondhi/README.md

Paras Sondhi

AI Engineer | Agentic Workflows & Production RAG

I build robust, backend-driven AI systems with a focus on multi-agent orchestration and retrieval-augmented generation. While my academic background is in Engineering at NIT Hamirpur, my technical execution is strictly focused on AI architecture, data science, and backend infrastructure.

🏆 Top 5.58% Nationally in GATE Data Science & AI (IIT Guwahati)

🛠 Core Stack

  • AI & Orchestration: LangGraph, LangChain, Groq, Gemini, Ollama (Local LLMs), HuggingFace
  • Backend & Data Processing: Python, FastAPI, Pandas, SQL
  • Data, Cloud & Deployment: ChromaDB, Docker, Render
  • Frontend: Streamlit

🚀 Featured Architecture

AutonomousResearchAgent | 🌐 Use Live App | ▶️ Watch Demo

  • Architecture: Deployed Human-in-the-Loop (HITL) multi-agent system built to iteratively research, synthesize, and automatically distribute reports.
  • Stack: LangGraph, Groq, Python, Render, Streamlit, SMTP
  • The Heavy Lifting: Successfully deployed the system to production using Render for the backend and Streamlit for the user interface. Built a Human-in-the-Loop (HITL) approval checkpoint using LangGraph, allowing users to intercept and refine the agent's search queries before execution. Engineered an automated SMTP integration to deliver the final compiled PDF report directly to the user's inbox.

Enterprise-RAG-Agent | ▶️ Watch Demo

  • Architecture: Zero-leakage hybrid infrastructure designed for 100% data privacy, capable of simultaneously querying unstructured PDFs and structured CSV/SQL databases locally.
  • Stack: Ollama (Local Execution), HuggingFace, ChromaDB, Pandas, SQL, LangChain, Python
  • The Heavy Lifting: Built a completely local, zero-data-leakage pipeline ensuring enterprise privacy using HuggingFace embeddings and Ollama. Engineered an automated data pipeline using Pandas to dynamically preprocess raw CSV files (handling missing values) before loading them into a SQL engine. Implemented vector storage (ChromaDB) for optimal PDF chunk retrieval, seamlessly routing user queries between the structured CSV data and the unstructured PDF text.

🌱 Open Source Contributions

  • py-goog-cli: Authored the core setup guide and identified/patched a CLI argument parsing error in the main documentation regarding model flag configurations.

📬 Let's Connect

Open to contract AI development work: RAG pipelines, agentic backends, LLM integrations, and production deployment.

Pinned Loading

  1. AutonomousResearchAgent AutonomousResearchAgent Public

    A live-deployed, asynchronous LangGraph agent that fully automates deep web research. Hosted on a decoupled Render backend, the agent executes recursive, self-correcting scraping loops, integrates …

    Python

  2. Enterprise-RAG-Agent Enterprise-RAG-Agent Public

    A 100% local, zero-leakage AI agent built with Llama 3, LangChain, and Streamlit to simultaneously query unstructured PDFs and structured SQL databases.

    Python