SQETCH 2026
Sqetch 2026 — Managed Quantum & QML Platform
Sqetch is Anzaetek's managed platform for quantum and quantum-machine-learning workloads. One entry point, multiple compatible hardware and simulation backends, federated training across sites, and confidential computing so data never leaves its source.
Sqetch '26 Finance is a development preview. A design-partner window is targeted for late 2026 — contact us to scope a workload.
- Deploy anywhere — cloud PoC, on-prem enclave, scale-out cluster
- Interfaces: Python SDK, REST, Excel; dockerized apps for backtests and QML
- Federated models across sites — data never leaves, only derived signals aggregate
- Multimodal QML — tabular, text, images, waveforms — with mid-circuit measurement for non-unitary models
Sqetch '26 Finance — HFT and beyond
- Worst-case risk management with quantum synthetic data (QCBM + Conditional TimeGAN)
- Market regime detection with Temporal Similarity Search over a time-series data lake
- Quantum optimal execution and portfolio optimization (QAOA / QUBO dispatched to the runtime)
- Hardware-accelerated backtests, script strategies, Q/ML-augmented repricing and XVA
OMEGA FUNCTIONS
OSS · Apache-2.0
Omega Functions — Open-Source Quantum Runtime
Omega Functions is our open-source Rust runtime for quantum and hybrid workloads. Think "AWS Lambda for quantum circuits" — write your circuit once, dispatch to the backend that fits the shape of your problem.
Released as open source on 2026-04-24 under Apache-2.0 at github.com/Anzaetek/omega-functions-public.
- Multi-backend: statevector, MPS tensor network, Pauli/Clifford, photonics (Gaussian)
- Adjoint differentiation for O(1) parameter gradients; CMA-ES for gradient-free loops
- Mid-circuit measurement, conditional gates, reset — across all backends
- WASM runtime (Wasmtime 28) for sandboxed classical control loops
- C ABI and REST server with PQC (ML-DSA-65) authentication
QUANTUM TOOLKIT
Quantum Toolkit — Circuit IR, QML & Formal Verification
Our internal quantum toolkit is a Rust programming environment on top of Omega Functions. It ships a circuit algebra, a QML stack with autograd, error-correction codes, and machine-checked proofs exported to Lean 4 and Rocq.
- 30+ gate types, symbolic parameters, circuit! proc-macro DSL, Aria surface DSL with resource estimation
- Format exporters — QASM 2.0, OPTICQASM 1.0, Lean 4, Rocq, JSON, .qc, omega-IR, C FFI
- 13 optimizer passes incl. PhasePolyDedup with F₂ Gaussian-elim kernel search and adjoint-pair cancellation
- QML with parameter-shift autograd through tch-rs; CMA-ES gradient-free; surrogate training (arXiv:2505.05249)
- Error-correction — repetition [[n,1,n]], Steane [[7,1,3]], surface code [[d²,1,d]] with MWPM decoder
- Formal verification — 23 standalone Lean 4 theorems (0 sorry); mathlib-backed ECC and BlockEncoding fully closed; QFT / QPE / Grover statements load-bearing, some packaging sorries open
The Quantum toolkit is closed source; the underlying Omega Functions runtime is open source.
QUANTUM-EXAMPLES
EVALUATION · RUN IT
quantum-examples — try the toolkit on real binaries
We publish a self-contained evaluation bundle of the Quantum toolkit at github.com/Anzaetek/quantum-examples. Thirteen demos run against the shipped binaries — circuit inspection and optimization, Aria→Lean 4 spec extraction, Quantum Oracle Sketching, blind delegated computation, error correction, and trained QML/finance models — with nothing on the host but a shell.
Evaluation-only: the binaries are time-limited and qubit-capped (12 general; 64 for the scalable ECC / Pauli-propagation paths). A commercial license is separate. Work in progress — expect rough edges.
- Self-contained binaries — no Rust, Python, or jq; verified end-to-end in a stock debian:bookworm-slim container
- macOS (Apple Silicon, Metal + CPU) and Linux x86-64 / arm64 (CPU) plus a linux-amd64 CUDA bundle that trains QML on Cuda(0)
- One-command guided tour (demo/run-all.sh) — each step self-checks against a numeric value
- All finance / QML data is synthetic — a rescaled, noised, renamed series, not real market data
Aria — the quantum language, now Apache-2.0 open source
The demo models are written in Aria, a small readable language with built-in proof obligations. It is now fully open source under Apache-2.0 at github.com/Anzaetek/aria-quantum-language-oss-public — a standalone pure-Rust repository with its own aria CLI and runtime that runs, trains, and exports from one source (OPENQASM 2.0, JSON, Lean 4). 31 of 32 examples are numerically verified against independent oracles, the Lean 4 proofs are sorry-free, and no libtorch is required (it is an opt-in backend). Editor tooling for VS Code, Neovim, and tree-sitter is included.