About us

Founded to solve a problem
seen first-hand

DeltaCode was founded in 2025 in Melbourne, Australia. Our mission is to close the gap between the code engineers write and the performance those engineers actually need.

The problem

Research code was never designed to be fast

Across scientific computing, quantitative finance, motorsport engineering and industrial simulation, the same pattern shows up again and again: a researcher or engineer writes a piece of numerical software that works. It gets used. Then it stays slow forever, because rewriting it is expensive and validating that a rewrite still produces the correct answer is even harder.

Engineering teams either spend significantly more on compute than necessary or defer optimisation work altogether because validating performance improvements is expensive.

DeltaCode was built to do this work automatically, and to prove it works using the same tests the codebase already trusts. If our optimisation cannot pass your existing tests end-to-end, we do not ship the pull request. Correctness first, always.

Our philosophy

How we think about optimisation

Three principles guide every optimisation DeltaCode delivers, and every product decision we make.

01

Correctness comes first

Optimisation is worthless if numerical results change. Every candidate is executed against your real inputs and must match the baseline output before it is ever proposed.

02

Evidence over promises

Every optimisation is benchmarked in a controlled container, with the full audit trail attached to the pull request. No claims about performance without the numbers to back them.

03

Engineers stay in control

DeltaCode proposes optimisations. Your engineers decide what gets merged. Every change goes through your normal pull request review before it touches your codebase.

Founder

Built by someone who lived the problem

Kyle Winkler
Kyle Winkler
Founder

Prior to founding DeltaCode, Kyle held simulation and performance engineering roles in Formula 1 and Australian Supercars, following research experience in quantum computing. Those experiences exposed a recurring challenge across engineering organisations: numerical software that was trusted in production but rarely revisited for performance optimisation.

Improving that software was rarely straightforward. Optimising critical numerical code often meant months of manual engineering effort, while proving that every optimisation preserved numerical correctness required equally rigorous validation. As a result, many codebases remained functionally correct but computationally inefficient, increasing infrastructure costs and slowing engineering workflows.

DeltaCode was founded to address that challenge. The platform automatically proposes performance improvements, validates every optimisation against a project’s existing test suite, benchmarks the results in an isolated environment, and generates a pull request for engineer review. The objective is simple: help engineering teams improve performance without compromising correctness or disrupting established development workflows.

Researchers and engineers should spend their time advancing their fields,
not rewriting working code for speed.

Our goal is to make high-performance scientific software accessible without months of manual optimisation. If that sounds like something your team could use, we’d like to talk.

Interested in working together?

Every engagement starts with a conversation. We'll walk through your codebase, understand what fast means to your team, and scope a first optimisation pass together.