Independent Research · Green Bay, WI

You can't
close source
the universe.

Robert Williams — independent researcher, systems thinker, guitarist.
Building post-quantum cryptography infrastructure on recycled hardware,
because migration shouldn't cost a fortune or wait until 2035.

Scroll
Robert Williams

Robert Williams · Jupiter Labs

Tinkerer first.
Researcher second.

I've been taking apart computers — and putting them back together — since I was a teenager. Started on Ubuntu and OpenSUSE around 2006, just because it was free and I was curious. That curiosity never really stopped.

During the 2017 crypto boom, GPUs went scarce and parents couldn't afford gaming PCs for their kids. So I started hitting Craigslist, sourcing parts — a GTX 960 here, an old i5 there — and cobbling together machines that could run Fortnite and Minecraft for $150–$200. Not fancy. Just functional. That was the whole point.

That same philosophy — use what's available, make it work, make it accessible — is exactly what drives CHIRASU today. The security industry says post-quantum migration is expensive and complicated. I say: not if you're creative about where you look.

I majored in astronomy before financial reality hit. I never graduated — but the way I think about systems, from subatomic noise to civilizational infrastructure, that's still very much an astrophysicist's lens. Every scale connects. Nothing is isolated.

By day I work as a parts coordinator. In the margins, I run a 20-node homelab, write post-quantum firmware in Rust and Go, play guitar in a band, and author a cyberpunk narrative universe that happens to share its philosophical DNA with my actual research. I don't separate these things. They're all the same inquiry.

2006–

First Linux install

Ubuntu, then OpenSUSE. I've been living in a terminal ever since. Still my daily driver on research hardware — Fedora and openSUSE depending on the machine.

2017–

The Craigslist era

Budget gaming rigs from salvaged parts — GTX 960s, old i5s, whatever Craigslist had. The goal was always to make something useful for people who couldn't afford the "right" solution. That instinct didn't go anywhere.

2022–

Aerospace Production

Manufacturing experience supporting programs including Artemis I and civilian and military aircraft. Working in environments where process rigor isn't optional teaches you something about what real validation actually looks like.

2025–

CHIRASU goes live

A 20+ node distributed PQC entropy harvesting system. Field-validated at a live concert on April 25, 2026. IACR preprint in progress.

The way I think
about everything.

Whether I'm debugging a Rust compile error on a single-board computer or thinking about post-quantum migration at civilizational scale, the same principle applies: zoom out far enough and the pattern repeats. Zoom in far enough and the noise becomes signal.

I keep coming back to the same question: the physical world is already generating randomness constantly — every motor load, every thermal drift, every photon hitting a sensor. That measurement is already happening, for other reasons. So why is "good entropy" something we treat as a service you subscribe to?

On entropy and trust

Services like Cloudflare's League of Entropy are genuinely useful — but they're also known, named, high-value targets. The moment your randomness source is a company with a URL, you've created a trust boundary an attacker can reason about. Distribute it. Make it physical. Make it local. Make it boring.

On recycled infrastructure

Old hardware is "adversarially boring." It's not on anyone's threat model spreadsheet. A 2012 Optiplex running Fedora, measuring grid fluctuation on its power supply, is producing legitimate entropy that no sophisticated attacker is accounting for. That's a feature, not a limitation.

On the 2035 myth

The industry narrative is that post-quantum migration is a decade away. It isn't. NIST has finalized ML-KEM and ML-DSA. The algorithms exist. The barrier is entropy quality and infrastructure inertia — both of which are solvable today, with creative sourcing and open tooling.

On closing source

True quantum randomness — QRNG — isn't a hardware module you buy. It's reality itself: shot noise in a camera sensor, RF interference, the exact thermal drift in your CPU at 3am. You can't patent that. You can't backdoor it. You can't close source the universe.

CHIRASU — Entropy Fusion

CHIRASU is a distributed post-quantum cryptography entropy harvesting system. The premise: civilization already instruments everything. Factory motor loads, satellite telemetry, IP camera feeds, temperature sensors, grid fluctuation — that measurement is already happening. CHIRASU harvests what's already being recorded, conditions it through NIST SP 800-90B health checks, and generates post-quantum keypairs at the edge.

It runs on a 20+ node mesh. Not a data center. Not cloud. A homelab of recycled Optiplexes, single-board computers, and guitar-mounted ESP32s — each one a feather node, passively observing its physical environment. The goal: a shared entropy pool distributed across the mesh. That architecture is actively in development.

ESP32 entropy harvester installed inside guitar body — GuitaRNG build GuitaRNG — ESP32 entropy harvester, mid-install

Hardware

x86 / Workstations
Dell Optiplex 3080 Dell Latitude series Dell Precision M5400 ROG Ally Z1 Extreme
Single-Board Computers
Radxa Rock 2A Radxa Zero 3W Libre Sweet Potato Libre Computer Alta Libre Renegade Libre Tritium H5 Orange Pi Zero 3 Orange Pi Zero 2W Raspberry Pi 5 Raspberry Pi Zero 2W
Microcontrollers
ESP32-S3 ESP32-C6 ESP32-P4+C6

Software

Primary languages
Rust Go Python Elixir / OTP Nim MicroPython Arduino / ESP-IDF Rust Embedded
Also used / experimental
Java Ruby Zig
Tooling
VSCode Notepad++ Thonny Headscale Cloudflared Forgejo

Operating Systems

Fedora openSUSE Armbian Debian Linux Mint Void Linux ParrotOS Windows 11
In progress — IACR ePrint target

CHIRASU: Harvesting Physical Entropy for Post-Quantum Key Generation at Scale

A four-tier decentralized architecture for entropy collection, conditioning, and PQC key generation from ambient physical sources — validated in field deployment at a live music event, April 2026.

Expected submission

July 2026

The constellation isn't just hardware collecting dust. These are active systems, experiments, and services running across the mesh right now.  ● active   experimental

Encrypted Ledger System

Hash-chained event ledger with a dedicated validator node. BLAKE3 for speed, SHA-256 for heavy-duty archival. Used to audit entropy events across the mesh.

PQC Key Generators

Two active generators producing ML-KEM-1024 and ML-DSA-87 keypairs from conditioned entropy, gated through NIST SP 800-90B health checks.

Self-Hosted Headscale

Full mesh VPN connecting all nodes. No Tailscale cloud dependency. Every device on the constellation reachable by hostname.

Multi-Source Entropy Harvesters

Active nodes pulling from piezo sensors, WiFi traffic deltas, Bluetooth, microphone noise, hardware RNG, and ADC pin noise across multiple devices simultaneously.

Self-Hosted NAS

Local archive and research storage. No cloud dependency for core data. Entropy bundles, keypair records, and field logs stored on-mesh.

CVE Puller + Audit

Automated vulnerability feed with confidence scoring and dependency auditing across the constellation's software stack.

Discord as Dashboard

Nearly all nodes post status, entropy readings, and alerts to a private Discord server via bots and slash commands. Low-overhead, always-on monitoring without a dedicated frontend.

Guitar Pickup Entropy

Harvesting entropy from ungrounded guitar pickups — the electromagnetic noise floor of an instrument at rest. Ongoing methodology experiment.

Camera / Photon Shot Noise

ISP-based adaptive gating on camera sensors to extract photon shot noise as a high-quality entropy source. Part of GuitaRNG Apex development.

The research doesn't
stay in the lab.

Two live deployments. Two different events. One through-line: working PQC entropy systems in uncontrolled, real-world environments. Not simulated. Not benchmarked in Docker. Actually in the room.

The Trax — Entropy Pipeline deployment

Entropy Pipeline · The Trax

First Live
Deployment

ESP32 entropy harvesters embedded in guitar hardware, streaming entropy during a live performance. The beginning of treating music as a research environment.

Missfits Tavern — PQC field validation, April 25 2026

PQC Field Validation · April 25, 2026 · Missfits Tavern

First end-to-end
pipeline test.

Null Cloak (mmWave radar), GuitaRNG Spectra and Sylvia (guitar-embedded ESP32s), and Null Magnet (laptop) — four devices, three entropy source types, running together in an uncontrolled environment. A proof of concept, not a production claim.

4 Devices deployed in field
3 Distinct entropy source types
426 ML-KEM-1024 keypairs generated
v0.1 Initial validation — whitepaper in progress

The work is a conversation,
not a monologue.

Independent research doesn't mean isolated research. These are some of the practitioners this work has connected with.

Standards Body

NIST / SP 800-90B

Early work reached the team responsible for NIST's entropy source validation standards. That conversation opened a path for post-publication dialogue and helped sharpen the technical framing considerably.

Quantum Security / Industry

CEO/CISO, Quantum Security Firm

A practitioner engaged seriously with the entropy pipeline approach and pushed back hard on trust boundary assumptions and propagation path visibility. That critique is now a load-bearing part of the whitepaper — exactly the kind of conversation independent research needs.

Academic Research / Industry

PhD Candidate, Major Semiconductor Firm

A formal research interview on practitioner PQC framing, conducted under academic ethics review. A useful calibration — the kind of feedback you can't get from reading papers alone.

If this resonates,
I'd love to hear from you.

I'm always open to collaboration, peer critique, research conversations, or just talking shop about entropy, embedded systems, or Linux distros. This work is more fun when it's a conversation.