The systems, and the reasoning behind them.

corerouter — a router whose every function we wrote ourselves

A complete home and edge router written from scratch in Rust — DHCP, DNS, firewall, multi-WAN failover, traffic shaping and a transactional config plane, running as native services on a board that costs less than a takeaway.

Status Pre-hardware-validation. Cross-built and verified by automated suites against a real kernel; not yet booted on a physical board.

NorthStar — end-to-end encrypted conferencing for regulated organisations

Browser-based video conferencing where the server never holds the keys — built for organisations whose legal and compliance teams need a better answer than "trust the vendor".

Status Pilot with design partners. Two-party encrypted calling verified end to end, including screen share; multi-party is in active development and is not yet verified as secure.

ChatForge — team messaging you host yourself

A self-hosted messaging platform for teams whose conversations should not live on someone else’s servers — real-time messaging, threads, presence, search and attachments, deployed on infrastructure you control.

Status Phase 1 — runs end to end and is tested. Encryption between participants, voice and video calling, push notifications and native mobile apps are deliberately out of scope for this phase.

Carrier Subscriber Edge with Closed-Loop Assurance

A broadband network gateway design that terminates subscriber sessions at scale, keeps convergence inside a hard budget, and streams the telemetry needed to detect degradation before the support queue does.

Line-Rate Packet Processing on Commodity Hardware

A userspace forwarding pipeline that reaches line rate on standard servers — poll-mode drivers, NUMA-aware placement, and a per-packet path engineered around the cache rather than around the code.

Fail-Safe OTA for a Distributed Device Fleet

An over-the-air update system for embedded Linux devices in the field: atomic A/B switching, cryptographic provenance from boot ROM to application, and a rollout that stops itself when the fleet starts failing.

Event-Driven Telemetry Platform for Industrial Assets

An ingestion and analytics platform for high-cardinality machine data — designed around backpressure, ordering guarantees and replay, so that a slow consumer degrades the system instead of losing data.