Skip to content
View ameeta18's full-sized avatar

Block or report ameeta18

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.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. 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
ameeta18/README.md

Hi, I'm Ameeta πŸ‘‹

Machine Learning & LLM Engineering Β· M.Sc. Computer Science

LinkedIn Email Location


About me

M.Sc. Computer Science student at RPTU Kaiserslautern focused on machine learning, LLM systems, and deep learning. Backed by 2+ years of production software engineering at Cognizant.

I work at the intersection of applied research and engineering: building multimodal deep learning pipelines, end-to-end RAG systems, and agentic LLM workflows. Comfortable moving between research rigor (designing architectures, running ablations, evaluating against strong baselines) and engineering pragmatism (writing tests, shipping through CI/CD, owning a task end-to-end).


πŸ”­ Featured Projects

A gated fusion architecture that predicts landslide risk from satellite data, designed to stay reliable when optical sensors fail during rain or cloud cover.

  • Precision 0.97 / Recall 0.83 on a geographically separated test set
  • Significantly outperformed Random Forest baseline (0.79 / 0.49)
  • Network autonomously learned the underlying physics: terrain gate 94.6%, moisture gate 74%
  • Zero spatial data leakage by design

PyTorch Rasterio GeoPandas NASA earthaccess Scikit-learn

End-to-end RAG system that automates academic research through agentic LLM orchestration.

  • 87% reduction in LLM context usage through chunking, embedding, and retrieval
  • 5Γ— more papers per session vs. naive context-stuffing
  • LangGraph ReAct agent orchestrating 6 custom LangChain tools (arXiv search, Semantic Scholar searc, PDF reader, vector indexer, semantic search, LaTeX renderer)
  • Custom evaluation framework for retrieval accuracy, faithfulness, and chunk coverage

LangGraph LangChain FAISS Google Gemini Streamlit Python FastAPI


πŸ› οΈ Tech Stack

Languages Python SQL JavaScript

ML & Deep Learning PyTorch TensorFlow scikit-learn NumPy Pandas

LLMs & GenAI LangChain LangGraph FAISS Hugging Face RAG

Frameworks & Tools FastAPI Streamlit Docker Git Azure DevOps


πŸ“ˆ GitHub Stats


πŸŽ“ Background

  • M.Sc. Computer Science β€” RPTU Kaiserslautern, Germany (Oct 2024 – Present)
  • B.Tech. Information Technology β€” RCC Institute of Information Technology (CGPA 8.75/10)
  • Programmer Analyst β€” Cognizant Technology Solutions (Aug 2022 – Sep 2024) Salesforce platform development, REST/SOAP API integration, Apex, Azure DevOps

Let's build something meaningful. πŸ“© ameetachaudhury18092000@gmail.com

Pinned Loading

  1. Landslide-Prediction-Project Landslide-Prediction-Project Public

    Multimodal deep learning and remote sensing framework for landslide susceptibility prediction

    Jupyter Notebook 1

  2. AI-Research-Assistant AI-Research-Assistant Public

    An intelligent research assistant that searches, analyzes, and writes academic papers using a Retrieval-Augmented Generation (RAG) pipeline. Built with LangGraph, FAISS vector database,MCP and Goog…

    Python 1

  3. Graph-Rag-Blast-Radius Graph-Rag-Blast-Radius Public

    a Graph RAG system for multi-hop dependency reasoning over microservice topologies

    Python

  4. multi-agent-code-review multi-agent-code-review Public

    A multi-agent code review system (GitHub Action) that reviews PRs with parallel security, maintainability, and test-coverage agents , benchmarked across Gemini, OpenAI, and Anthropic with a hand-la…

    Python