Skip to content
View ShubhamSongire's full-sized avatar

Highlights

  • Pro

Block or report ShubhamSongire

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.

Please don't include any personal information such as legal names or email addresses. Maximum 100 characters, markdown supported. This note will be visible to only you.
Report abuse

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

Report abuse
ShubhamSongire/README.md

Hi πŸ‘‹, I'm Shubham Songire

AI/ML Engineer | RAG Systems | M.Tech AIML @ BITS Pilani

Ex: DRDO Β· IDfy Β· Mass Technologies

Machine Learning Engineer | Generative AI | LangChain | NLP | Computer Vision | AI Research | ROS

shubhamsongire

shubhamsongire


πŸš€ About Me

  • πŸ”­ I’m currently working on Retrieval-Augmented Generation (RAG) systems for clinical trial data using OpenSearch, LangChain, Flask, Docker & AWS
  • 🧠 Pursuing M.Tech in Artificial Intelligence & Machine Learning from BITS Pilani
  • πŸ› οΈ Building Video-to-Video Summarization pipeline with transcripts, narration & screenshot generation
  • πŸ§ͺ Exploring LLMs, multi-modal AI, and prompt engineering
  • πŸ’¬ Ask me about NLP, LLM pipelines, MLOps, or backend APIs
  • πŸ“« Reach me at: [email protected]

πŸ“Œ Featured Projects

  • 🧠 Clinical RAG System
    Built a scalable Retrieval-Augmented Generation system using OpenSearch + S3 + LangChain, with evaluation via RAGAs, deployed on AWS via Docker.

  • πŸ“Ί YouTube Video Summarizer
    Extracts transcripts & screenshots β†’ narrates + generates summary video. Includes interactive Q&A using transcript-based retrieval.

  • 🧬 Biomedical NER & Sentence Classification
    Developed NLP models for entity recognition & sentence intent classification using HuggingFace, PyTorch, and spaCy.


πŸ”— Connect with Me

Twitter


Skills:

Programming Languages:

python matlab matlab c cplusplus java

Generative AI:

langchain langgraph langsmith ollama openai gemini chroma huggingface

Libraries and Frameworks:

numpy pandas matplotlib sklearn sns selenium OpenCV MediaPipe

Web Development:

streamlit flask fast-api html5 css3 javascript postman

Cloud and Deployment:

google-cloud-platform vertex-ai amazon-web-services docker vercel

Databases:

firebase Oracle-SQL MySQL Python-SQLite Sanity-CMS

Tools and Platforms:

linux excel git github vsc

Hardware:

raspberrypi arduino



πŸ“ˆ GitHub Stats

Shubham's GitHub Stats

GitHub Streak


πŸ† Certifications

  • βœ… Azure AI Fundamentals (AI-900)
  • πŸ“Š Applied Data Science Specialization – Coursera

Feel free to ⭐ any repo you find interesting!

Pinned Loading

  1. YouTubeVideoToVideoSummarizer-QA YouTubeVideoToVideoSummarizer-QA Public

    Intelligent Video Knowledge Extraction: YouTubeVideoToVideoSummarizer is a powerful tool that transforms YouTube videos into queryable knowledge bases using advanced Retrieval Augmented Generation …

    Python 1

  2. Graph-Summary-Generator Graph-Summary-Generator Public

    The system effectively classifies different types of graphs using a CNN model, and writes short summary about graph in a clear and concise paragraph-like format.

    Jupyter Notebook 1

  3. Sky-Fleet-Classifier-using-CNN-Imagenet-models Sky-Fleet-Classifier-using-CNN-Imagenet-models Public

    I have classified drone images dataset using some imagenet models. I got better accuracy from ResNet152V2 model.

    Jupyter Notebook 1

  4. Contact-Tracing-with-Machine-Learning Contact-Tracing-with-Machine-Learning Public

    Contact tracing is the process of identifying and tracking individuals who have come into contact with an infected person. This method is used by governments to help curb the spread of infectious d…

    Jupyter Notebook 1 1

  5. Identify_unreliable_news_articles_using_NLP Identify_unreliable_news_articles_using_NLP Public

    This project uses NLP and machine learning to identify fake news from a dataset on a Kaggle competition, providing a robust approach to classify and stop the spread of false information.

    Jupyter Notebook 1

  6. COVID-19-Chest-x-ray-challenge COVID-19-Chest-x-ray-challenge Public

    The following is the premise for this challenge: Available COVID-19 chest x-ray data from a given facility is limited. This is like the mini-dataset. Unseen images may come from various sources tha…

    Jupyter Notebook 1