Nadeem Sajjad – Senior AI Engineer · Forward-Deployed Engineer

Overview

NewsFlow is a news collaboration platform that transforms traditional RSS feeds into an Instagram-style scrollable reel experience. Instead of scrolling through dense article lists, users pick a channel and swipe through full-bleed, image-rich news cards, one story at a time.

The platform currently supports English (SCMP) and Hindi (Dainik Bhaskar) channels with category filtering. A major upcoming feature will translate every headline and summary into the reader's preferred spoken language: Hinglish, regional dialects, and colloquial forms, making global news genuinely accessible to everyone.

Key Features

Channel Selection: browse and pick from multiple RSS-based news channels (SCMP, Dainik Bhaskar, more coming).

Reel-Style Feed: full-screen, swipeable news cards with snap-scroll, just like Instagram reels.

Category Filtering: switch between World, Tech, Business, Sport, and more within each channel.

Image-Rich Cards: each story features a hero image, headline, summary, source attribution, and share button.

Keyboard Navigation: arrow keys for desktop swiping through stories.

RSS Parsing Pipeline: server-side XML parsing with channel-specific parsers for clean, structured data.

Regional Language Translation (coming): auto-translate headlines and summaries into spoken languages like Hinglish and colloquial dialects.

Mobile-First Design: full-viewport cards, touch-friendly swipe, and responsive layouts built for phones first.

How It Works

The landing page presents a grid of available channels, each with a logo, name, and default category. When a user selects a channel, the app navigates to a full-screen reel view. An API route fetches the channel's RSS feed server-side, parses it with a channel-specific parser, and returns structured JSON with headlines, summaries, images, authors, and timestamps.

The frontend renders each story as a full-viewport card with a background image, gradient overlay, headline, and summary. Users swipe vertically (or press arrow keys on desktop) to move between stories. A progress indicator on the right tracks position, and a swipe hint appears on first load. Each card includes a share button and a direct link to the original article.

The upcoming translation layer will intercept the parsed feed data and route it through a language model to rewrite headlines and summaries in the user's chosen dialect, not just official languages, but the way people actually speak. Hinglish, regional slang, and colloquial forms will all be supported.

Challenges & Solutions

Challenge

RSS feeds from different publishers have wildly different XML structures and content quality.

Solution

Built a channel-specific parser system. Each channel gets its own parser that normalizes data into a unified NewsItem type, so the frontend always receives consistent, predictable data regardless of the source.

Challenge

Creating a smooth, Instagram-like vertical scroll experience on the web.

Solution

Used CSS snap-scroll (snap-y snap-mandatory) with full-viewport cards and dvh units for true full-screen slides. Combined with touch-action and passive scroll listeners for native-feeling swipe without jank.

Challenge

Handling feed caching and freshness without hammering external RSS servers.

Solution

Used Next.js revalidation (5-minute cache) on the server-side fetch so RSS requests are cached and only refreshed periodically, keeping response times fast while staying reasonably up-to-date.

Challenge

Translating news into colloquial and spoken languages rather than formal regional dialects.

Solution

Planned a translation pipeline that routes parsed content through a language model fine-tuned on informal speech patterns. Instead of Hindi → Hindi or English → English, it transforms to Hinglish, brotherhood slang, and other spoken forms that match how people actually communicate.

Outcome

NewsFlow reimagines news consumption for the mobile-first era, replacing dense article grids with an engaging, swipeable reel format. By pulling from RSS feeds and rendering full-bleed story cards, it makes keeping up with the news feel as effortless as scrolling social media.

The project is actively in development. The core reel experience and RSS parsing pipeline are live, with the regional language translation layer as the next major milestone. Once launched, it will offer a genuinely new way to consume news, in the language you actually speak.