Atlas — Distributed Infrastructure Engine

Infrastructure is modelled as a graph — BFS, DFS, Dijkstra and A* pathfinding, topological sort, cycle detection, connected components, Tarjan SCCs, articulation points and bridges, across directed and undirected representations.

A from-scratch library covering dynamic arrays, singly and doubly linked lists, stacks, queues, circular queues, binary min-heaps, priority queues, chaining hash maps, BST and AVL trees, tries and an LRU cache — with an in-process benchmark harness timing all 16 workloads and recording runtime, memory and throughput.

Resource-aware scheduling with a transparent score breakdown (CPU, memory, GPU, network, health and affinity, plus a utilisation penalty and priority bonus), and a priority-based job queue with retries.

Leader election, log replication, heartbeats, term management and snapshots implemented directly. A replicated key–value store commits writes through the Raft log and serves reads from a consistent view.

A hash ring with virtual nodes places shards and keeps re-balancing minimal when the node set changes.

kill_node, network_partition, latency, cpu_stress and memory_pressure experiments with automatic revert and blast-radius impact analysis — wired into the event bus so failures flow naturally into anomaly and incident systems.

Rolling-window z-score detection plus absolute threshold rules (defaults: CPU > 90, memory usage > 85) feed an incident manager that coalesces related events, correlates across the cluster and drives a lifecycle from open to investigating to resolved.

Rule-based archetype analysis — deliberate failure injection, dependency cascade, single-component failure, cluster-wide pressure, capacity bottleneck, latency degradation. Advisory output is read-only and every recommendation requires explicit approval before anything is acted on.

Prometheus text and JSON metrics endpoints, event counters, structured logging, an operational CLI (atlas-cli) and a Next.js dark-theme dashboard showing topology, services, open incidents, anomalies and active chaos.