Capstone Projects
Build production-grade distributed systems from scratch. Each project is a deep dive into a real-world system with multiple tracks, subtracks, and tasks.
Mini-Dynamo
Implement Amazon Dynamo from the 2007 paper. Build a fully functional distributed key-value store featuring consistent hashing, quorum replication, sloppy quorums, hinted handoff, vector clock conflict detection, Merkle tree anti-entropy, and read repair.
Mini-Kafka
Build a production-grade message broker from scratch. Implement Kafka's append-only partition log, Raft-based leader election, ISR tracking and high-watermark advancement, idempotent and transactional producers, and consumer groups with rebalancing.
Mini-Spanner
Build a globally consistent distributed database from scratch. Implement Paxos consensus, TrueTime external consistency, two-phase commit across shards, MVCC snapshot reads, and a SQL query router.