Stare at the abyss long enough, and it starts to stare back at you...
Working notes from building backends — mostly distributed systems, the infrastructure under them, and the maths a few of them depend on. Written to be re-read later, by me.
The failure modes that only appear under concurrency, retries and real traffic — and that pass review because they work fine in development. Grouped by the subsystem they bite: runtime, storage, transactions, data modelling, APIs, distribution, and payments.
43 pages
Running MicroK8s on one bare-metal node with no public IP: a NetBird overlay for reachability, Nginx Proxy Manager in front, per-person RBAC certificates, S3 backups, and a rebuild checklist for when it all has to come back from nothing.
12 pages
WebSocket gateway clusters built to fan out exchange events at low latency, MPC crypto-custody over NATS with authenticated encryption, recurring billing, and the SDD template the rest are written against.
8 pages
Elliptic curves from the group law upward — what ECC actually computes, how ECDSA signs with it, why secp256k1 picks the parameters it does, and the modular inverse the whole thing rests on.
4 pages
rust-libp2p in practice: the transport and muxing stack, the protocols everything else is built on, NAT traversal when neither peer is addressable, and a decentralised p2p playground to try it in.
4 pages
Why agreement across replicas is hard, and two protocols that manage it anyway — Raft, which elects a leader everyone trusts, and Viewstamped Replication, which changes view when that leader stops answering.
2 pages
Also hereInterview prep4Reference4Trading1Browse by tag121Archive4