Production software for systems with real failure cost
Production software · real failure cost
Trebuchet Dynamics builds production software for scientific compute, infrastructure, validation engines, GPU systems, and agent runtimes. Founder-led delivery. No offshoring.
We build systems where failure has real operational cost, and we design accordingly: legal identity, source ownership, delivery evidence, and a clear engagement model kept ready for procurement review.
Public repositories, shipped apps, and source-backed research tools make the engineering audit trail inspectable before procurement. Six shipped public systems are available to inspect. Arenatón is in development, Protein Statera Go is PRD-ready, and additional production systems remain under NDA.
Runtime and market infrastructure where deployability, signing boundaries, and operator recovery matter.
Go-native agent runtime with operational gateway work for dependable installs, diagnostics, and recoverable sessions.
Single Go binary and SDK for production-safe Polymarket V2 trading with local deposit-wallet signing boundaries.
Research tools and source-backed engines built for repeatable validation, reporting, and review.
Desktop research stack for FeCIM simulation workflows, references, and validation harnesses.
Source-backed nuclear-data engine for isotope records and auditable alpha-decay traversal.
Cosmology and relativity toolkit for residual-first model comparison against observed datasets.
Go workbench plan for confidence, geometry, RMSD, and evidence-bound protein reports.
Shipped and in-progress user surfaces that turn specialized systems into usable products.
Published GPU fractal explorer with deep zoom, high-resolution export, and no ads or tracking.
Web and mobile interface for the public 9lives.so prediction-markets protocol on Superposition.
Led by Juan Manuel (principal engineer, building production systems since 2015) with 3 in-house engineers. Senior review stays close to the work: architecture, implementation, demos, and source handoff stay connected. The relevant background is physics, mathematics, Go, Rust, and C# applied to lower-risk handoff for systems with real failure cost.
Senior engineering stays close to architecture, code, demos, and handoff.
Weekly checkpoints, working software, green CI, and explicit priorities.
Source, docs, runbooks, and release process move toward a lower-risk handoff from day one.
Audit the artifact trail before signing: repositories, shipped apps, diagnostics, and source-backed reports.
Engagements run from 4-week sprints (~£40k) to multi-year builds (£250k+). We work in fixed-scope phases, not hourly billing. Rate cards shared after a discovery call.
1–2 weeks, fixed-price. You leave with a signed technical spec, risk inventory, and a fixed-price scope estimate for the build phase. No commitment to proceed.
Yes to both. Source, docs, and runbooks are owned by you from day one. NDAs, DPAs, MSAs, and full-confidentiality engagements are routine — we adapt to your procurement.
Continuity is the studio's responsibility, not yours. The founder leads every engagement; in-house engineers shadow critical paths. If anyone leaves, we backfill at our cost — no schedule slip charged to you.
Send the problem, constraints, timeline, and decision process. We'll respond with whether the engagement model is the right next step.