Full-stack engineer architecting media streaming pipelines, resilient SaaS backends, and AI systems that scale under real-world production load.
I’m Indal Singh, a full-stack engineer who specializes in the high-concurrency layers of products — distributed media transcoding pipelines, real-time audio loopbacks, and multi-tenant SaaS platforms that hold up under heavy production traffic.
I began programming in 2018 and stepped into professional software engineering in 2022. Starting with independent client contracts, I quickly advanced to owning full-stack production architecture at Play A Sport — engineering high-throughput booking engines, Redis cache invalidation, and automated CI/CD container pipelines.
My engineering practice bridges systems-level programming (C++20 WASAPI loopbacks) with modern cloud backends (Node.js, BullMQ, Docker). I’m currently open to senior full-stack roles, media engineering challenges, and technical consulting engagements.
From building distributed media processing microservices to directing end-to-end web architecture for production platforms across Delhi, Mumbai, and Kolkata.
Leading end-to-end engineering for the Play A Sport web platform. Responsible for designing scalable backend systems, database architecture & query optimization, DevOps deployment pipelines, cloud infrastructure, and enterprise-grade security.
Direct architectural ownership of Play A Sport's web platform — engineering the high-throughput booking engine, Redis caching layers, and multi-tier RBAC security serving sports academies across India.
Architecting scalable RESTful microservices and APIs powering the user-facing booking portal and multi-tenant academy management SaaS.
Designing relational & document data models, complex query optimization, indexing strategies, and transactional integrity for slot & batch bookings.
Managing cloud environments on AWS, Cloudflare Zero Trust perimeter defenses, Docker containerization, CI/CD automated deployment pipelines, Nginx reverse proxies, and 99.9% uptime monitoring.
Enforcing Cloudflare Zero Trust access policies, multi-tier Role-Based Access Control (RBAC) across users, coaches, and academy owners, JWT session hardening, rate limiting, and secure payments.
The missing macOS 'Multi-Output Device' natively built for Windows 10 & 11. Plays system audio simultaneously through multiple Bluetooth earbuds, USB headsets, and speakers with millisecond sync delay and zero virtual drivers.
High-performance YouTube music streaming PWA with screen-off background playback via MediaSession API, 90% mobile data and battery savings through pure Opus/AAC audio extraction, real-time Web Audio API spectrum visualizer, and 320kbps MP3 downloader.
Scalable video processing microservice built with BullMQ and Redis queues. Orchestrates multi-bitrate HLS adaptive transcoding via FFmpeg, generates preview thumbnails, and uploads packaged segments directly to AWS S3 with worker concurrency controls.
Real-time streaming server that ingests raw binary media chunks over WebSockets from browser clients, pipes them through a low-latency FFmpeg transcoding pipeline, and relays broadcasts to RTMP live endpoints with real-time viewer tracking.
Production problem breakdowns across distributed media processing, live streaming gateways, and low-latency audio engineering — backed by verifiable metrics.
Monolithic upload servers hung on 2GB+ raw video uploads (MP4/MOV) and lacked adaptive bitrate packaging, causing 8–15s buffering and playback failures on spotty mobile networks.
Engineered an asynchronous distributed queue using BullMQ and Redis with per-worker concurrency throttling. Automated FFmpeg multi-rendition segmentation (1080p to 360p) with direct multipart S3 streaming and thumbnail extraction.
Processed 2GB+ files with zero web server blocking, reduced CDN video bandwidth consumption by ~45%, and slashed first-frame playback startup time to under 3 seconds.
Web clients needed direct live video broadcasting from mobile browsers to RTMP ingests (YouTube Live/Twitch). Traditional RTMP required desktop OBS Studio, while existing web solutions suffered 3–5s latency and high dropped-frame rates.
Implemented a Node.js WebSocket streaming gateway that ingests binary MediaRecorder chunks, pipes them via child-process stdin directly into FFmpeg (x264 ultrafast, zero-latency tuning), and broadcasts real-time FLV over TCP.
Achieved sub-second ~850ms glass-to-glass latency from standard mobile browsers at 60fps 1080p, sustaining concurrent live sessions with under 120MB server memory footprint per stream.
Windows natively restricts playback to a single active audio endpoint, preventing dual-headphone sharing. Furthermore, hardware DSP differences across Bluetooth earbuds introduce a 40–120ms delay discrepancy that creates intolerable acoustic echo.
Engineered a native C++20 WASAPI loopback capture engine with lock-free Single-Producer Single-Consumer (SPSC) ring buffers, coupled with independent 0–500ms delay compensation lines and zero virtual driver dependencies.
Eliminated kernel driver installs entirely (0 drivers required), slashed memory usage to <25 MB RAM, and achieved sub-5ms internal audio engine processing latency with 1ms precision per-device sync.
Type a command and get a quick response about my work, stack, projects, or contact details.
Welcome to indal.dev — interactive portfolio terminal v1.0.0
Type `help` to list commands, or try: about · experience · skills · projects · contact
Open to roles and projects in media infrastructure, streaming, and AI systems.
