Fractional CTO and engineering leadership. Architecture review, technical due diligence, and interim head-of-engineering work, mostly for teams with hard systems problems.

The repos here are public work product. All three are MIT licensed. Architecture field notes reviewing them — what they do, where the boundaries sit, and what the results do and do not establish — are published at arboreng.com/field-notes.

github.com/arboreng/cislunar-sim · Field note

Spacecraft trajectory simulation from LEO through cislunar space, in Python. A validated physics engine plus Q-law guidance as an importable library — not an application driven by mission scripts, like GMAT, or a JVM hosted from Python, like Orekit.

The force model covers J2/J3 and 8-mascon GRAIL lunar gravity, third-body solar perturbation, Hall-effect ion thrust, solar sail SRP, NRLMSISE-00 drag, eclipse shadowing, and quaternion attitude dynamics, integrated with adaptive RK4(5). Runtime dependencies are numpy, astropy, and nrlmsise00.

Validation against LightSail 2:

pip install cislunar-sim

github.com/arboreng/pq-migration-lab

Post-quantum cryptography migration patterns, in Go. The algorithms are standardized and reference implementations exist; what's less documented is what breaks when a running system moves to them.

Four migration demos, each runnable end to end:

• hybrid-tls — TLS 1.3 handshake over X25519MLKEM768 with no fallback group offered on either side
• cert-migration — one classical Ed25519 CA issuing both Ed25519 and ML-DSA-65 leaf certificates
• ca-rotation — rotating a CA from Ed25519 to ML-DSA-65 with a cross-signed transitional cert
• interop-tls — the same hybrid group between Go's crypto/tls and OpenSSL with oqs-provider, in both directions

Benchmarks for X25519 vs. ML-KEM-768 and Ed25519 vs. ML-DSA-65 are generated from go test output.

docker compose run pqlab demo hybrid-tls

github.com/arboreng/quantum-foundry

Quantum algorithm implementations in Python on Qiskit. Public implementations of these algorithms are usually toy demonstrations or framework examples; what's missing is the scaffolding around them — tests, benchmarks, and a written path from the mathematics to the circuit to the code.

Nine algorithms, each with a test suite, a benchmark harness, and its mathematics and circuit derivation written out alongside the implementation:

• Query and promise problems — Deutsch-Jozsa, Bernstein-Vazirani, Simon
• Phase and period finding — Quantum Phase Estimation, Shor (also covering gate-decomposed arithmetic and hardware-aware transpilation)
• Search — Grover
• Variational — QAOA, VQE
• Linear systems — HHL

Each one is checked end to end against its published result — Shor against Nielsen & Chuang's N=15, Simon against the 1994 hidden period, HHL against Harrow-Hassidim-Lloyd (2009) — through the algorithm's public API rather than its internals. All nine pass from a clean clone. The suite is 296 tests with a 92% coverage floor.

A cross-algorithm study transpiles a comparable circuit from every algorithm onto the same linear nearest-neighbor coupling map, and finds that multi-controlled gates rather than qubit count drive routing cost: Simon's 10-qubit circuit routes with zero swaps, while HHL's 5-qubit circuit needs 50.

Every algorithm currently sits at experimental on the repository's maturity model, and hardware-behavior results are not yet part of it.

uv run python -m validation.known_answers

arboreng.com · sean@arboreng.com