A Deep Dive Into the Architecture of an AI-Powered News Summarization App, NewsGPT (Neotice), and Deploying It on E2E Cloud

January 15, 2024


In today’s digital age, the way we consume news has undergone a significant transformation. The persistent wave of information and the abundant clickbait culture pose a considerable challenge for those seeking quick and reliable news. Consumers are constantly bombarded with headlines that vie for attention, often at the expense of factual accuracy. This information overload makes it difficult to filter through the noise and find content that is both trustworthy and relevant. Even few reputed media houses use clickbaits and fake news to attract readers to their site.

NewsGPT (Neotice) is a new open source application created by Tim K.C. Ho, designed to cut through the clutter and deliver news that matters, and was one of the applications that won the Streamlit Hackathon. 

Summary of The Technologies Used 

NewsGPT uses LlamaIndex, Apache Spark, Firebase and Qdrant Vector DB to not only summarize news articles but also ensure the information is pulled from credible sources. The NewsGPT github repo explains how to deploy and build on it.

Apache Spark

Apache Spark™ is a highly regarded open-source, distributed processing system designed for large-scale data workloads. Spark is recognized for its ability to handle batch and real-time analytics and data processing tasks. It unifies these capabilities within a single framework, making it a go-to solution for a wide range of data engineering, data science, and machine learning applications on both single-node machines and clusters. 

Apache Spark includes several components that enhance its utility across various domains:

Spark Core: This is the foundation of the platform, handling memory management, fault recovery, job scheduling, and other fundamental functions​​.

Spark SQL: A distributed query engine that provides low-latency, interactive queries. It's used for processing structured data with SQL or Hive Query Language, making it suitable for a range of applications from business analytics to data reporting​​.

Spark Streaming: This component is designed for real-time data processing. It allows the processing of live data streams, making it a robust solution for real-time analytics​​.

MLlib (Machine Learning Library): Spark includes a library of machine learning algorithms, enabling scalable and efficient data science applications. It supports various tasks like classification, regression, and clustering​​.

GraphX: A library for graph processing, enabling users to interactively build and transform graph data structures at scale​


LlamaIndex is a comprehensive data framework specifically designed for enhancing Large Language Model (LLM) applications. It acts as a bridge between custom data sources and LLMs, enabling users to integrate, manage, and query a wide range of data types using natural language.

Key Features of LlamaIndex

Data Ingestion: LlamaIndex supports a variety of data sources, including APIs, PDFs, documents, SQL, and more, allowing users to connect their existing data formats with LLM applications​​.

Data Indexing: It provides capabilities to store and index data for various use cases. This involves converting data into a format that can be easily queried and understood by LLMs​​​​.

Query Interface: One of the standout features is its query interface, which allows for knowledge-augmented responses to input prompts over the indexed data. This enables the generation of responses that are contextually relevant and informed by the underlying data​​​​.

How LlamaIndex Works

Ingestion: LlamaIndex can connect custom data sources to LLMs using hundreds of data loaders. It supports a wide range of pre-built solutions like Airtable and Salesforce, as well as generic plugins for loading data from various formats​​.

Indexing: After ingestion, data is indexed to be mathematically represented, allowing for easy querying by LLMs. Different types of indexing models, such as vector, list, tree, and keyword indexes, cater to specific application needs​​.

Querying: The querying process in LlamaIndex is intuitive, using natural language through prompt engineering. This simplifies the interaction with data, making it more accessible and user-friendly​​.


Firebase is a comprehensive backend platform, offering a suite of cloud-based tools and services for application development. 

Key Features and Services of Firebase

Authentication: Firebase provides a robust authentication system supporting various methods, including email, phone numbers, and social media platforms like Google, Facebook, and Twitter. It also offers FirebaseUI for building familiar sign-in user interfaces​​.

Real-Time Database and Cloud Firestore: These are NoSQL databases that offer real-time data syncing across client devices. The Real-time Database stores and syncs data in a JSON-like format, whereas Firestore organizes data into collections and documents for more complex, hierarchical data structures​​.

Hosting: Firebase offers easy-to-configure hosting for websites, supporting static frontends and dynamic content via technologies like HTML, CSS, JavaScript, and Express.js. It provides automatic HTTPS configuration and fast delivery through a global Content Delivery Network​​.

Cloud Functions: This is a serverless framework allowing developers to write functions in Node.js, which respond to various events triggered within Firebase or via HTTP requests. It's designed for high scalability and reliability​​.

Cloud Storage: Suitable for storing user-generated content like images and videos, Firebase Storage ensures robust, resilient uploads and downloads​​.

Cloud Messaging: This service enables the sending of notifications to users to enhance engagement and user retention

Qdrant Vector DB

Qdrant is an open-source vector database and vector search engine, predominantly written in Rust. It is designed to offer fast and scalable vector similarity search services with a user-friendly API. Qdrant facilitates the efficient handling of high-dimensional vectors, making it particularly suitable for applications that involve embeddings or neural network encoders. These capabilities enable Qdrant to support a wide range of use cases, including matching, searching, recommending, and much more in data-intensive applications.

One of Qdrant's strengths lies in its ability to manage embeddings produced by neural network encoders, which are essential for semantic-based matching and various other applications. It supports a range of features such as faceted search, making it versatile for different kinds of data matching and retrieval tasks. The platform is also designed to be easy to set up and use, offering different methods for deployment including Docker images and a Python client, as well as a cloud-based solution for production environments.

Deploying NewsGPT on E2E Cloud

NewsGPT is open source, so you can easily customize and deploy it for your own individual use case. To do so, head to E2E Cloud, sign up, and launch a GPU node. 

E2E Cloud has A100 Cloud GPU Server, and has recently also launched HGX 8xH100, the most powerful AI supercomputer currently available, for instant access. 

Once you have launched your GPU node, clone the repository, and set up your Python virtual environment, and install the requirements. 

source /bin/activate
git clone https://github.com/timho102003/NewsGPT
cd NewsGPT
pip install -r requirements.txt
python Home.py

The current repo has been setup with the following in the config, which you can modify:


This mentions the news domains, and the number of articles to fetch for the feed.

How It Works

The technological framework of NewsGPT is built upon real-time news aggregation, and then a RAG pipeline that creates embeddings from each of the articles, and allows the LLMs to query and access the information within the articles.

Data Pipeline

The data pipeline of NewsGPT starts with Google News, which acts as a primary source for the latest news articles across various domains. This is followed by asynchronous web scraping, a process designed to collect data efficiently without being bottlenecked by the sequential loading of articles. 

The scraped content is then channeled through Kafka, a distributed event streaming platform that handles the flow of data with high throughput and scalability. 

From Kafka, topics are organized and funneled into Spark Streaming, a technology adept at processing real-time data streams. It performs two critical functions here: it recognizes named entities within the content, and it generates embeddings. Named Entity Recognition (NER) isolates and categorizes key components from the text, such as individuals, organizations, and locations.

The generation of embeddings translates articles into vectorized forms that capture the semantic essence of the content. These vectors are then stored in the Qdrant Vector DB, which serves as a repository for the processed data, ready for retrieval and analysis.

News Flow - Nivas Jayaseelan

Personalization and Recommendation

Personalization is at the heart of NewsGPT, where user preferences and activities dictate the content that is recommended. When a user interacts with the platform, their choices and behaviors are captured—what articles they read, what topics they linger on, and what they search for.

This data is funneled into the Qdrant Vector Database, where it informs the recommendation algorithms. The database acts as a matchmaker, aligning the user's interests with articles from a vast pool of content. The result is a tailored news feed that feels personal because it is shaped by the individual's own reading patterns and preferences.

Chat with Articles

The Chat with Articles feature is where the LlamaIndex Retriever-Augmented Generation (RAG) system shines. This system engages with the user's query by searching through the indexed vectors in the Qdrant Vector Database for related articles. 

It then synthesizes information from these articles to generate a comprehensive and concise response. This is where the app becomes interactive, as users can engage in a dialogue with the system, probing deeper into topics of interest or asking for clarifications, much like they would with a human expert. 

A screenshot of a computerDescription automatically generated

Each query triggers a cascade of events within the LlamaIndex RAG, which pulls together various sources, summarizes, compares, and contrasts information to provide a response that is informative and nuanced. The responses are then presented through the Chat Bot Display, offering users a seamless and engaging experience.

Streamlit Workflow - Nivas Jayaseelan

The flowchart provided helps visualize the web of processes that make up the NewsGPT platform. Each component is interconnected from data sourcing to user interaction, creating a harmonious and intelligent system for news consumption.

Core Ideas behind NewsGPT

  • Reliable News Sources: NewsGPT's dynamic data pipeline is the backbone of its operation, beginning with a sophisticated algorithm that sources news from credible outlets. This system ensures that the information being summarized is accurate and up-to-date. The application employs advanced Named-Entity Recognition (NER) to identify and categorize key information from articles, while text embedding processes, facilitated by state-of-the-art machine learning techniques, capture the essence of the content. This is all meticulously organized within a Qdrant Vector Database, ensuring that the news is not only reliable but also relevant and ready for rapid retrieval.
  • Tailored News Recommendations: Understanding that each user's interests are unique, NewsGPT personalizes the news-reading experience through a learning algorithm that adapts to individual reading habits. By analyzing engagement with different topics and stories, the app refines its news recommendations, creating a bespoke news feed that becomes more aligned with the user's preferences over time. This personalized approach is reflected in the flowchart's depiction of user interaction with the vector database, where preferences and activities inform the recommendation engine.
  • Efficient Information Retrieval: NewsGPT simplifies the process of obtaining comprehensive news summaries through a one-click solution. By aggregating and summarizing news from multiple reliable sources, the app provides a multifaceted view of current events. This feature is powered by a Streamlit chatbot, which presents a conversational interface for users to interact with, allowing for an engaging and efficient retrieval of news summaries. The flowchart illustrates how, upon a user's query, the chatbot interfaces with the LlamaIndex RAG to pull responses from a vector store of indexed articles, delivering concise summaries directly to the user.
  • Time-Saving Aspects: NewsGPT provides quick summaries. It features a sidebar that displays the amount of time users would have spent reading full articles versus the time taken to read the summaries, offering a tangible sense of efficiency. Moreover, the app provides insights into news category distribution, allowing users to visualize the spread of their news consumption across different topics, as depicted in the flowchart through the various streams of article recommendation and retrieval.

Each of these features comes together to create an app that not only respects the intelligence and time of its users but also enhances their ability to stay informed in an era of information saturation.

Why Build an AI Platform for News?

The consumption of news has seen a profound evolution over the years, driven by technological advancements and changing consumer preferences. From the days when reading the newspaper was a morning ritual, we have journeyed through an era where radio and television brought the world's events into our living rooms, creating a shared experience for families nationwide. 

The advent of the 24-hour news cycle and cable networks catered to an ever-growing demand for up-to-the-minute news, while the rise of the internet brought about a seismic shift, as news websites and online editions of newspapers provided real-time updates and a platform for interactive engagement.

This evolution continued with the emergence of social media and blogs, which used news distribution, allowing anyone with internet access to both consume and share information. The mobile revolution and app-based news consumption personalized news delivery, with alerts and feeds tailored to individual preferences. 

Now, news aggregators and summarization services are the latest type of news, which condenses information from various sources. Despite the changing landscape, the public's need for reliable, factual news has remained constant, with the latest technologies striving to balance the demand for brief, accessible content with the essential need for accuracy and integrity.

Timeline of News Media - Nivas Jayaseelan

Credible news sources actually do more than report facts; they provide context, depth, and analysis, helping readers understand the complex world around them. They are the antidote to the misinformation spread by clickbait, offering a more nuanced and comprehensive view of events. 

In a time where there is a lot of information, but insight is scarce, tools that can differentiate through the noise and present news from reliable outlets are not just convenient; they are necessary for an informed public. 

This is where RAG-pipeline driven AI comes in, an approach where contextual data is fetched from credible sources, and thereby generating summaries that are more informed and accurate. NewsGPT uses Apache Spark for Named Entity Recognition, and capabilities of LlamaIndex to filter news articles into concise, digestible summaries from trusted sources. 

In the coming future, AI would become part of the workflow in news and media industries.

The Future of News and Conclusion

The future of news is currently at the exciting intersection of technology and information. NewsGPT points to a possible new future direction that technology would possibly take in the news industry, offering a powerful mixture of speed, accuracy, and personalization. Its impact is not just in how news is consumed, but in how it is experienced, turning the passive act of reading into an active dialogue. 

The efficiency and responsiveness of AI-driven platforms are heavily reliant on robust computational power. This is where cloud GPU services that provide instant access to advanced GPUs come into play, offering scalable resources that can process large datasets, manage machine learning tasks, and handle user queries in real-time. In India, the company that pioneered advanced cloud GPUs is E2E Networks, offering the most advanced AI Supercomputer HGX 8xH100 to developers and researchers. The way to get started is by heading to MyAccount and signing up. 

On E2E Cloud, developers can choose from a range of GPUs, including the powerful A100, H100 and HGX 8xH100 at a nominal price. If you're ready to start building or expanding your AI-driven solutions, you can get started today by signing up. Moreover, E2E Cloud offers a wide variety of other GPUs to suit different needs and budgets, ensuring that you can find the perfect fit for your project.


Latest Blogs
This is a decorative image for: A Complete Guide To Customer Acquisition For Startups
October 18, 2022

A Complete Guide To Customer Acquisition For Startups

Any business is enlivened by its customers. Therefore, a strategy to constantly bring in new clients is an ongoing requirement. In this regard, having a proper customer acquisition strategy can be of great importance.

So, if you are just starting your business, or planning to expand it, read on to learn more about this concept.

The problem with customer acquisition

As an organization, when working in a diverse and competitive market like India, you need to have a well-defined customer acquisition strategy to attain success. However, this is where most startups struggle. Now, you may have a great product or service, but if you are not in the right place targeting the right demographic, you are not likely to get the results you want.

To resolve this, typically, companies invest, but if that is not channelized properly, it will be futile.

So, the best way out of this dilemma is to have a clear customer acquisition strategy in place.

How can you create the ideal customer acquisition strategy for your business?

  • Define what your goals are

You need to define your goals so that you can meet the revenue expectations you have for the current fiscal year. You need to find a value for the metrics –

  • MRR – Monthly recurring revenue, which tells you all the income that can be generated from all your income channels.
  • CLV – Customer lifetime value tells you how much a customer is willing to spend on your business during your mutual relationship duration.  
  • CAC – Customer acquisition costs, which tells how much your organization needs to spend to acquire customers constantly.
  • Churn rate – It tells you the rate at which customers stop doing business.

All these metrics tell you how well you will be able to grow your business and revenue.

  • Identify your ideal customers

You need to understand who your current customers are and who your target customers are. Once you are aware of your customer base, you can focus your energies in that direction and get the maximum sale of your products or services. You can also understand what your customers require through various analytics and markers and address them to leverage your products/services towards them.

  • Choose your channels for customer acquisition

How will you acquire customers who will eventually tell at what scale and at what rate you need to expand your business? You could market and sell your products on social media channels like Instagram, Facebook and YouTube, or invest in paid marketing like Google Ads. You need to develop a unique strategy for each of these channels. 

  • Communicate with your customers

If you know exactly what your customers have in mind, then you will be able to develop your customer strategy with a clear perspective in mind. You can do it through surveys or customer opinion forms, email contact forms, blog posts and social media posts. After that, you just need to measure the analytics, clearly understand the insights, and improve your strategy accordingly.

Combining these strategies with your long-term business plan will bring results. However, there will be challenges on the way, where you need to adapt as per the requirements to make the most of it. At the same time, introducing new technologies like AI and ML can also solve such issues easily. To learn more about the use of AI and ML and how they are transforming businesses, keep referring to the blog section of E2E Networks.

Reference Links




This is a decorative image for: Constructing 3D objects through Deep Learning
October 18, 2022

Image-based 3D Object Reconstruction State-of-the-Art and trends in the Deep Learning Era

3D reconstruction is one of the most complex issues of deep learning systems. There have been multiple types of research in this field, and almost everything has been tried on it — computer vision, computer graphics and machine learning, but to no avail. However, that has resulted in CNN or convolutional neural networks foraying into this field, which has yielded some success.

The Main Objective of the 3D Object Reconstruction

Developing this deep learning technology aims to infer the shape of 3D objects from 2D images. So, to conduct the experiment, you need the following:

  • Highly calibrated cameras that take a photograph of the image from various angles.
  • Large training datasets can predict the geometry of the object whose 3D image reconstruction needs to be done. These datasets can be collected from a database of images, or they can be collected and sampled from a video.

By using the apparatus and datasets, you will be able to proceed with the 3D reconstruction from 2D datasets.

State-of-the-art Technology Used by the Datasets for the Reconstruction of 3D Objects

The technology used for this purpose needs to stick to the following parameters:

  • Input

Training with the help of one or multiple RGB images, where the segmentation of the 3D ground truth needs to be done. It could be one image, multiple images or even a video stream.

The testing will also be done on the same parameters, which will also help to create a uniform, cluttered background, or both.

  • Output

The volumetric output will be done in both high and low resolution, and the surface output will be generated through parameterisation, template deformation and point cloud. Moreover, the direct and intermediate outputs will be calculated this way.

  • Network architecture used

The architecture used in training is 3D-VAE-GAN, which has an encoder and a decoder, with TL-Net and conditional GAN. At the same time, the testing architecture is 3D-VAE, which has an encoder and a decoder.

  • Training used

The degree of supervision used in 2D vs 3D supervision, weak supervision along with loss functions have to be included in this system. The training procedure is adversarial training with joint 2D and 3D embeddings. Also, the network architecture is extremely important for the speed and processing quality of the output images.

  • Practical applications and use cases

Volumetric representations and surface representations can do the reconstruction. Powerful computer systems need to be used for reconstruction.

Given below are some of the places where 3D Object Reconstruction Deep Learning Systems are used:

  • 3D reconstruction technology can be used in the Police Department for drawing the faces of criminals whose images have been procured from a crime site where their faces are not completely revealed.
  • It can be used for re-modelling ruins at ancient architectural sites. The rubble or the debris stubs of structures can be used to recreate the entire building structure and get an idea of how it looked in the past.
  • They can be used in plastic surgery where the organs, face, limbs or any other portion of the body has been damaged and needs to be rebuilt.
  • It can be used in airport security, where concealed shapes can be used for guessing whether a person is armed or is carrying explosives or not.
  • It can also help in completing DNA sequences.

So, if you are planning to implement this technology, then you can rent the required infrastructure from E2E Networks and avoid investing in it. And if you plan to learn more about such topics, then keep a tab on the blog section of the website

Reference Links



This is a decorative image for: Comprehensive Guide to Deep Q-Learning for Data Science Enthusiasts
October 18, 2022

A Comprehensive Guide To Deep Q-Learning For Data Science Enthusiasts

For all data science enthusiasts who would love to dig deep, we have composed a write-up about Q-Learning specifically for you all. Deep Q-Learning and Reinforcement learning (RL) are extremely popular these days. These two data science methodologies use Python libraries like TensorFlow 2 and openAI’s Gym environment.

So, read on to know more.

What is Deep Q-Learning?

Deep Q-Learning utilizes the principles of Q-learning, but instead of using the Q-table, it uses the neural network. The algorithm of deep Q-Learning uses the states as input and the optimal Q-value of every action possible as the output. The agent gathers and stores all the previous experiences in the memory of the trained tuple in the following order:

State> Next state> Action> Reward

The neural network training stability increases using a random batch of previous data by using the experience replay. Experience replay also means the previous experiences stocking, and the target network uses it for training and calculation of the Q-network and the predicted Q-Value. This neural network uses openAI Gym, which is provided by taxi-v3 environments.

Now, any understanding of Deep Q-Learning   is incomplete without talking about Reinforcement Learning.

What is Reinforcement Learning?

Reinforcement is a subsection of ML. This part of ML is related to the action in which an environmental agent participates in a reward-based system and uses Reinforcement Learning to maximize the rewards. Reinforcement Learning is a different technique from unsupervised learning or supervised learning because it does not require a supervised input/output pair. The number of corrections is also less, so it is a highly efficient technique.

Now, the understanding of reinforcement learning is incomplete without knowing about Markov Decision Process (MDP). MDP is involved with each state that has been presented in the results of the environment, derived from the state previously there. The information which composes both states is gathered and transferred to the decision process. The task of the chosen agent is to maximize the awards. The MDP optimizes the actions and helps construct the optimal policy.

For developing the MDP, you need to follow the Q-Learning Algorithm, which is an extremely important part of data science and machine learning.

What is Q-Learning Algorithm?

The process of Q-Learning is important for understanding the data from scratch. It involves defining the parameters, choosing the actions from the current state and also choosing the actions from the previous state and then developing a Q-table for maximizing the results or output rewards.

The 4 steps that are involved in Q-Learning:

  1. Initializing parameters – The RL (reinforcement learning) model learns the set of actions that the agent requires in the state, environment and time.
  2. Identifying current state – The model stores the prior records for optimal action definition for maximizing the results. For acting in the present state, the state needs to be identified and perform an action combination for it.
  3. Choosing the optimal action set and gaining the relevant experience – A Q-table is generated from the data with a set of specific states and actions, and the weight of this data is calculated for updating the Q-Table to the following step.
  4. Updating Q-table rewards and next state determination – After the relevant experience is gained and agents start getting environmental records. The reward amplitude helps to present the subsequent step.  

In case the Q-table size is huge, then the generation of the model is a time-consuming process. This situation requires Deep Q-learning.

Hopefully, this write-up has provided an outline of Deep Q-Learning and its related concepts. If you wish to learn more about such topics, then keep a tab on the blog section of the E2E Networks website.

Reference Links



This is a decorative image for: GAUDI: A Neural Architect for Immersive 3D Scene Generation
October 13, 2022

GAUDI: A Neural Architect for Immersive 3D Scene Generation

The evolution of artificial intelligence in the past decade has been staggering, and now the focus is shifting towards AI and ML systems to understand and generate 3D spaces. As a result, there has been extensive research on manipulating 3D generative models. In this regard, Apple’s AI and ML scientists have developed GAUDI, a method specifically for this job.

An introduction to GAUDI

The GAUDI 3D immersive technique founders named it after the famous architect Antoni Gaudi. This AI model takes the help of a camera pose decoder, which enables it to guess the possible camera angles of a scene. Hence, the decoder then makes it possible to predict the 3D canvas from almost every angle.

What does GAUDI do?

GAUDI can perform multiple functions –

  • The extensions of these generative models have a tremendous effect on ML and computer vision. Pragmatically, such models are highly useful. They are applied in model-based reinforcement learning and planning world models, SLAM is s, or 3D content creation.
  • Generative modelling for 3D objects has been used for generating scenes using graf, pigan, and gsn, which incorporate a GAN (Generative Adversarial Network). The generator codes radiance fields exclusively. Using the 3D space in the scene along with the camera pose generates the 3D image from that point. This point has a density scalar and RGB value for that specific point in 3D space. This can be done from a 2D camera view. It does this by imposing 3D datasets on those 2D shots. It isolates various objects and scenes and combines them to render a new scene altogether.
  • GAUDI also removes GANs pathologies like mode collapse and improved GAN.
  • GAUDI also uses this to train data on a canonical coordinate system. You can compare it by looking at the trajectory of the scenes.

How is GAUDI applied to the content?

The steps of application for GAUDI have been given below:

  • Each trajectory is created, which consists of a sequence of posed images (These images are from a 3D scene) encoded into a latent representation. This representation which has a radiance field or what we refer to as the 3D scene and the camera path is created in a disentangled way. The results are interpreted as free parameters. The problem is optimized by and formulation of a reconstruction objective.
  • This simple training process is then scaled to trajectories, thousands of them creating a large number of views. The model samples the radiance fields totally from the previous distribution that the model has learned.
  • The scenes are thus synthesized by interpolation within the hidden space.
  • The scaling of 3D scenes generates many scenes that contain thousands of images. During training, there is no issue related to canonical orientation or mode collapse.
  • A novel de-noising optimization technique is used to find hidden representations that collaborate in modelling the camera poses and the radiance field to create multiple datasets with state-of-the-art performance in generating 3D scenes by building a setup that uses images and text.

To conclude, GAUDI has more capabilities and can also be used for sampling various images and video datasets. Furthermore, this will make a foray into AR (augmented reality) and VR (virtual reality). With GAUDI in hand, the sky is only the limit in the field of media creation. So, if you enjoy reading about the latest development in the field of AI and ML, then keep a tab on the blog section of the E2E Networks website.

Reference Links




Build on the most powerful infrastructure cloud

A vector illustration of a tech city using latest cloud technologies & infrastructure