Gen AI Document Summarization and QA

I am a data scientist with around 4 years of relevant experience and extensive knowledge of Web Scraping, Python, NumPy, Pandas, PyTorch, Keras, sci-kit-learn, Generative AI and NLP. I have successfully implemented lot of scraping websites with different complexities from various websites such as e-commerce, music websites and so on. I have also worked on projects like a Multi-Doc QA system using OpenAI embeddings and GPT3 and retrieving relevant questions and answers to a user's query using OpenAI embeddings deployed on Azure App services for multiple concurrent users within a timeline of 1 week. I have experience working on different Generative AI tools such as GPT3.5/GPT4, llama2-7b in delivering industry grade projects. I have extensive experience with Building Generative AI Applications, Prompt Engineering, Azure Cloud Platform and LLM Models. I can develop streamlit application for generating summaries of user uploaded documents using LLM Models, Langchain and Map Reduce techniques within 7 working days. I can assist you in deployment of the application as well, if needed. I am dedicated to delivering high-quality work that exceeds expectations, and I believe my skills and experience make me a great fit for your project. If you have any concerns related to the budget of this project then we can discuss in the chats. I will be happy to consider your inputs. Happy to connect with you and discuss freely in the chat regarding the project details and deadlines!

Hello, I checked Gen AI Document Summarization and QA is a good project to work on, with 10+ years of experience in development we are ready to start and deliver it on time, we can disucess the details in chat. Thanks

Hi, I’ve worked on a similar project where I built a pipeline that generates embeddings from uploaded documents, stores them in FAISS/ChromaDB, and uses RAG for answering queries. This solution is live in production. For your project, I’ll use ibm-granite/granite-3.3-2b-instruct on Google Colab Free with HuggingFace and Langchain to: Generate embeddings and store them in FAISS Use RAG for QA Implement map-reduce summarization for queries with ["summary", "summarize", "brief", "overview"] Build a Streamlit UI with file upload and an interactive text box Return answers in clean text or Markdown I have strong experience with Colab, Langchain, and building production-ready AI workflows. Looking forward to working together!

I propose a lightweight, scalable AI solution to summarize and query banking documents like policies, notices, and guidelines. The system will use the ibm-granite/granite-3.3-2b-instruct model via Hugging Face on Google Colab (Free tier), with LangChain to manage prompt workflows and RAG-based logic for contextual QA. The app will support keyword-triggered summarization using terms like “summary,” “summarize,” “overview,” and “brief,” leveraging a map-reduce pipeline. For retrieval, FAISS will be used to build vector indexes, ensuring fast, accurate responses even with large datasets. The front end will be built using Streamlit, offering an intuitive UI with file upload, an interactive text box, and output in Markdown/Text for improved user experience. With over 17 years in AI/ML and 5+ years in Generative AI, I specialize in deploying LLM-based solutions using LangChain, FAISS, and vector databases. I’m confident in delivering a robust solution within 5–7 working days. Looking forward to collaborating and transforming document intelligence through AI. Rohit Gupta Founder – Hustle AI Solutions

With hands-on experience in building scalable LLM-powered tools at Webuters Technologies, I bring a strong foundation in: Semantic Search & QA: Built a PDF Bot using LangChain and GPT-4o-mini for secured/unsecured files. Summarization Pipelines: Delivered a document summarizer for raw text, URLs, and PDFs. Custom AI Tools: Created CodeGenie and Resume Analyzer using best practices in LLM app development. LLM Integration: Deep familiarity with LangChain, vector databases (Pinecone, FAISS), and Colab deployment constraints. I have build similar projects under the same and can deliver to the best of client's needs on time.

Having spent years honing my skills in AI and ML, I consider the potential of leveraging Hugging Face for your Gen AI Document Summarization and QA to be nothing short of exciting. If selected for this project, I will draw upon my extensive experience working with Generative AI, IBM models, HuggingFace, and Langchain Frameworks to provide you with a top-notch solution that meets all your requirements. I am also deeply knowledgeable in the use of FAISS, RAG for QA-related purposes. One area where I truly excel and can bring significant value to your project is my strong background in map-reduce techniques. Implementing efficient map-reduce for summarization whenever user input query has keywords like ["Summary","Summarize","Brief","Overview"] is something I'm particularly skilled at. Additionally, I can create a Streamlit UI that not only includes necessary functionalities like file upload and an interactive text box but is also expertly designed for easy UX. One area where I truly excel and can bring significant value to your project is my strong background in map-reduce techniques. Implementing efficient map-reduce for summarization whenever user input query has keywords like ["Summary","Summarize","Brief","Overview"] is something I'm particularly skilled at. Additionally, I can create a Streamlit UI that not only includes necessary functionalities like file upload and an interactive text box but is also expertly designed for easy UX.

I've strong experience and hands on leveraging generative AI for various simple to complex tasks including chatbot creating and deploying them to streamlit UI.

I will build a Gen AI-powered Streamlit app on Google Colab Free using IBM Granite 3.3-2B-Instruct via HuggingFace and Langchain. It will summarize banking documents using map-reduce when triggered by specific keywords, and perform QA using FAISS and RAG. The app will include file upload and a smart interactive textbox for input. Output will be in clean text/Markdown for better UX. I have strong experience with Colab, Streamlit, FAISS, and Langchain, ensuring a smooth and efficient solution.