Company

A shared foundation for financial intelligence.

We start with the research workflow: data, models, evaluation, and monitored paper deployment. Our ambition is to turn that foundation into a market for strategies and infrastructure for portfolio construction.

Contact the founder →

Founder

Built by someone who has built this class of system before.

TSTim Shoaf · Founder & CEO
Tim Shoaf
Founder & CEO

Tim built, scaled, and led the ML Platform & Infrastructure organization at Roblox, the systems that train, serve, and monitor nearly every model at the company, and stood up the data platform behind an 80PB+ data lake ingesting tens of billions of events a day. His work sits at the intersection of high-performance computing, distributed systems, and machine learning: the disciplines Promethean is made of.

Promethean is the platform he wanted and could not buy: one system, written in Rust, that owns the path from raw data to a deployed, reproducible piece of financial intelligence, together with the market and portfolio layer that a system like that makes possible.

Where we are

Status, as of September 2026.

live

The development platform end to end: data, features, experiments, backtesting with provenance, registry, deployment to monitored paper sessions on real market data, metering and billing, multi-tenant workspaces, experiment comparison and live monitoring, in-app learning.

building

The marketplace: evaluation records and discovery are live; access for capital and the economics around it are being structured.

direction

The portfolio layer: composition infrastructure in design; conversations with allocators and managers underway.

later

Real-money execution and any offering to investors come after the regulatory and venue work they require.


Thesis

A shared foundation for independent research.

For decades, quantitative intelligence was produced inside a small number of institutions with the infrastructure, data, and capital to do it properly. That is ending: the tools of machine learning are widely held, the data is increasingly accessible, and the people capable of producing specialized intelligence far outnumber the funds that can employ them.

We are building the layer that lets that intelligence be built rigorously, evaluated comparably, distributed to capital, and composed into products. We believe that layer is the durable position, more durable than any single strategy or fund, for three reasons:

  • Evidence is the scarce asset. Strategies are abundant; strategies whose behaviour can be trusted are rare. Producing trustworthy evidence at scale is an infrastructure problem.
  • Markets form around comparability. Once evidence is standardized, a market in intelligence becomes possible, and it draws builders and capital to each other.
  • Composition could compound. With sufficient participation and permission to use the evidence, a shared evaluation environment could reveal how independently built strategies behave together.

The next test is adoption: researchers who use the platform repeatedly and value the evidence it preserves.

Business sequence

Start with the research workflow.

Today: platform access and usage. Managed research workspaces, metered compute and storage, and platform billing provide the starting commercial foundation. Researchers can use the platform for their own work before participating in the marketplace.

Next: distribution. Listings and evaluation records are the first step. Access for professional allocators and builder economics depend on the partner and regulatory work now being structured.

Longer term: portfolio infrastructure. Comparable strategy histories could support evaluation and construction at the portfolio level. That opportunity depends on adoption, evidence quality, and the rights to use that evidence. It remains a product direction.

Team

Small on purpose.

TS
Tim Shoaf
Founder & CEO

Led the ML Platform & Infrastructure organization at Roblox; built the data platform behind an 80PB+ lake. HPC, distributed systems, machine learning.

AS
Amber Sharma
Product

Owns the surfaces where the platform meets people, translating a deeply technical engine into a product builders and allocators want to use. Background in operations and analysis, with formal grounding in data privacy (CIPP/E, CIPP/US) and hands-on machine-learning training.

Principles

Five commitments.

01

Reproducible, or it did not happen

A result counts when it can be re-derived.

02

Correctness in the data path

Point-in-time, leakage, and lineage are properties of the system.

03

Infrastructure before market

The market is only as good as its evidence; the evidence is only as good as the platform.

04

Rust at the core

Business logic in a memory-safe, concurrent engine; Python as a declaration.

05

Say what is live

Every status on this site is true at the time of writing.

What we are looking for

Help build the next stage.

Quant

Quantitative research lead

Owns evaluation methodology, backtest fidelity, and the standard a listing must meet.

Engineering

Distributed systems, in Rust

The execution and data planes: the runtime that turns a declaration into a trained, evaluated, deployed piece of intelligence.

Interested?

We are small on purpose and talking to people directly. If one of these is you, write to tim@cryptex.ai and say what you would own.


Where do you enter?

For company investment, partnerships, or a walkthrough of the platform, contact Tim directly.

Talk to the founder