Longer notes on the systems I've built, 2025 – 2026.
Real-time online coding platform
An online IDE where multiple users edit and run code on the same project in real time, supporting up to 100 concurrent users. Each project gets its own isolated Kubernetes pod, with ingress controllers routing project-specific requests and WebSockets linking services for live collaboration.
MERN · Docker · Kubernetes · WebSockets
Fault-tolerant encoding pipeline
A master-worker video encoding pipeline that runs in parallel across machines, using gRPC for task scheduling and failure recovery. Raft-inspired leader election and heartbeats keep tasks coordinated when the master node fails, and FFmpeg splits streams into keyframe-aware segments to make full use of available compute.
Python · gRPC · Raft · FFmpeg · Kubernetes
Replicated telemetry store
A distributed storage platform in Python and C++ that ingests high-volume telemetry data and replicates it across nodes, so reads keep being served when a node goes down. Node-to-node sync is built on TCP/IP sockets with batched updates to keep replicas consistent, and a real-time dashboard shows node health, replication lag, and storage usage.
Python · C++ · TCP/IP