TranscodeX
A distributed video processing and streaming platform featuring multipart uploads, FFmpeg-powered HLS transcoding, AI-generated captions, transcript-based chat, realtime processing updates, and adaptive bitrate streaming.
Overview
Transcodex is a full-stack video processing platform inspired by modern streaming systems such as YouTube. The project implements the complete media pipeline from upload to playback — including multipart uploads, distributed video processing, thumbnail generation, HLS transcoding, adaptive streaming, AI-generated captions, transcript extraction, realtime progress tracking, and transcript-aware conversational AI. The primary goal was to explore media infrastructure, asynchronous processing, realtime communication, and AI integration in a single production-style system.
What Users Can Do
- Upload large video files using multipart uploads directly to AWS S3.
- Track video processing progress in realtime without polling.
- Watch videos using adaptive bitrate HLS streaming.
- Manually switch between 480p, 720p, and 1080p playback quality.
- Control playback speed and view generated subtitles.
- Access automatically generated captions and transcripts.
- Chat with an AI assistant that answers questions using the video's transcript as context.
- View thumbnails generated automatically during processing.
Why I built this
- To understand how modern video platforms process and deliver media at scale.
- To gain hands-on experience with FFmpeg, HLS, and adaptive streaming.
- To learn distributed job processing using BullMQ and Redis.
- To implement realtime communication using Socket.IO and Redis Pub/Sub.
- To explore practical AI integrations beyond simple chatbot applications.
- To build a portfolio project demonstrating backend engineering, cloud infrastructure, media processing, and AI in a single system.
Tech Stack
After launch & Impact
- Implemented multipart uploads to AWS S3 for reliable large-file handling.
- Built a distributed processing pipeline using BullMQ workers and Redis queues.
- Created an FFmpeg-based transcoding system that generates multi-resolution HLS streams.
- Implemented adaptive streaming with manual quality selection and playback controls.
- Built realtime processing updates using Socket.IO and Redis Pub/Sub.
- Integrated Whisper-based caption generation and transcript extraction.
- Implemented transcript-aware AI chat using Gemini.
- Designed a media pipeline that separates API responsibilities from processing workloads.
Future Plans
- Implement semantic transcript search using vector embeddings.
- Add timeline-aware AI responses that link answers to video timestamps.
- Generate AI-powered video summaries and chapter markers.
- Introduce GPU-accelerated transcoding for faster processing.
- Add CDN integration for global video delivery.
- Deploy horizontally scalable worker pools for parallel media processing.