DY Research is independent technical intelligence for investors, founders and engineering teams. Every claim a technology makes is traced to its code, its tests and its evidence, and the engagement ends in a written verdict on what holds.
Don't tell me the claim is correct. Give me the evidence: what is assumed, what is proven, what fails, and where the model stops being valid.
The tooling is open source. The verdict is yours.
| Engagement | The question it answers | Price |
|---|---|---|
| Snapshot · 5 business days | What does this technology actually do, and what does its evidence support? | $5,000 |
| Focused Audit · 2–3 weeks | Does one critical property — timing, determinism, concurrency — actually hold? | $12,000 |
| Due Diligence · 3–4 weeks | Is the technology what the company says it is, and what could break the investment? | $25,000 |
Fixed price and scope, confirmed in writing before any work begins; invoiced in USD or EUR. Every engagement is carried out by the principal, start to finish. Engagements and full scope → · the scope document · connect@axonos.org
A response-time bound is only as good as what is fed to it and what it is used for. The timing work is four tools, one per question, each doing its part in integer arithmetic and refusing where it cannot justify an answer.
dy-trace → dy-blocking → dy-wcet → dy-certify
execution blocking response a certificate an
times terms times auditor can re-run
| Tool | What it produces | Status |
|---|---|---|
| dy-wcet | Worst-case response times, exactly, with a named refusal wherever a bound cannot be justified | Shipped · 4.1.6 · 8 Kani proofs and 100 tests in CI · 0 dependencies |
| dy-trace | Execution-time inputs with their confidence stated | In design |
| dy-blocking | Blocking terms derived from a resource graph, not estimated | In design |
| dy-certify | A signed schedulability certificate an auditor can re-run | In design |
A tool in design has a README that says what it will do and a section on what is not there yet. Nothing in design is claimed to work.
Two tasks. A runs 100 µs every 400 µs at higher priority; B runs 200 µs every 1,000 µs. The common answer for B is 400 µs. The correct one is 300 µs — and with other periods, the same mistake reports a deadline as met that is missed on hardware.
The schedule, drawn by simulating the scheduler rather than the formula. On 5,000 random task pairs it lands exactly on the analysis's answer.
Try it live → · Source · The method · The bounty
The method, applied to the practice's own tool: this is what every engagement produces for every claim that matters.
| Step | For dy-wcet |
|---|---|
| Claim | Response times are computed exactly, in integer arithmetic |
| Source | Stated as a design rule in the README |
| Code | u64 throughout, and no floating point — checked by audit.sh on every push |
| Tests | 100 tests, fifteen of them derived by hand |
| Proof | 8 Kani proofs, each closing in CI, each bounded over the ranges it declares |
| Verdict | Evidenced, within its stated scope |
| An RP2350 timer that stopped for minutes | An intermittent embassy-time failure traced through alarm arming and timer-queue liveness, with evidence and hypothesis kept apart. The standard of delivery for a Focused Audit |
| 557 confident wrong answers | The same review, run on dy-wcet itself, found it had reported 557 task sets as meeting deadlines they miss. Found, fixed, and published in full |
| Two implementations, one answer | A second response-time analysis, written separately, agrees on a set where the fixed point is not the first value tried |
Revenue from every engagement funds AxonOS — an open-source, deterministic systems layer for neurotechnology — and its path to independent foundation governance. The fee pays for two things: the independent technical intelligence you receive, and open infrastructure for the field.
Funding AxonOS never shapes a verdict. If a company under review competes with AxonOS or builds on it, you are told at scoping, before you commit to anything.
|
Build — AxonOS |
Measure — dy-wcet |
|
Discover — Radar |
Verify — DY Research |
The founder's account, with the live map and the demos: AxonOS-BCI.
- Not investment advice. The findings are technical; the decision stays yours.
- Not a certification. dy-wcet is not a qualified tool under any safety standard, and no engagement issues or implies a standard's qualification.
- Not a warranty. What you receive is evidence and reasoning, set out so that it can be checked.
The tools are Apache-2.0 OR MIT, at your option. The engagements are written agreements.
© DY Research / Denis Yermakou
dyresearch.github.io · connect@axonos.org · LinkedIn · AxonOS
