<?xml version="1.0" encoding="UTF-8"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>SpeyTech - Technical Articles</title><description>Technical articles on deterministic computing, safety-critical systems, and production AI architecture.</description><link>https://speytech.com/</link><language>en-us</language><copyright>Copyright 2026 SpeyTech</copyright><managingEditor>william@fstopify.com (William Murray)</managingEditor><webMaster>william@fstopify.com (William Murray)</webMaster><atom:link href="https://speytech.com/rss.xml" rel="self" type="application/rss+xml"/><image><url>https://speytech.com/favicon.svg</url><title>SpeyTech</title><link>https://speytech.com/</link></image><item><title>Incident Reconstruction: Why &apos;It Worked Yesterday&apos; Isn&apos;t Evidence</title><link>https://speytech.com/ai-architecture/incident-reconstruction-determinism/</link><guid isPermaLink="true">https://speytech.com/ai-architecture/incident-reconstruction-determinism/</guid><description>How bit-perfect replay, execution tracing, and sealed audit logs transform incident response from guesswork to forensics</description><pubDate>Mon, 26 Jan 2026 22:00:00 GMT</pubDate><category>AI Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>When Fixed-Point Beats Floating-Point (And When It Doesn&apos;t)</title><link>https://speytech.com/insights/fixed-point-vs-floating-point-tradeoffs/</link><guid isPermaLink="true">https://speytech.com/insights/fixed-point-vs-floating-point-tradeoffs/</guid><description>An honest analysis of when Q16.16 is the right choice and when floating-point is acceptable</description><pubDate>Mon, 26 Jan 2026 21:30:00 GMT</pubDate><category>Deterministic Computing</category><author>william@fstopify.com (William Murray)</author></item><item><title>Version Control for Deterministic Systems: Git Isn&apos;t Enough</title><link>https://speytech.com/ai-architecture/version-control-deterministic-systems/</link><guid isPermaLink="true">https://speytech.com/ai-architecture/version-control-deterministic-systems/</guid><description>How Merkle chains, cryptographic attestation, and reproducible builds satisfy certification evidence requirements</description><pubDate>Mon, 26 Jan 2026 21:15:00 GMT</pubDate><category>AI Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>The Init-Update-Status-Reset Pattern: O(1) Guarantees for Safety Monitors</title><link>https://speytech.com/insights/init-update-status-reset-pattern/</link><guid isPermaLink="true">https://speytech.com/insights/init-update-status-reset-pattern/</guid><description>A four-function interface that enables static analysis, bounded resources, and compositional verification</description><pubDate>Mon, 26 Jan 2026 20:05:00 GMT</pubDate><category>Systems Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>Contracts as Documentation: Why Comments Lie and Code Doesn&apos;t</title><link>https://speytech.com/insights/contracts-as-documentation/</link><guid isPermaLink="true">https://speytech.com/insights/contracts-as-documentation/</guid><description>How preconditions, postconditions, and invariants become living documentation</description><pubDate>Mon, 26 Jan 2026 19:00:00 GMT</pubDate><category>Formal Methods</category><author>william@fstopify.com (William Murray)</author></item><item><title>Why &apos;Hello World&apos; Fails Safety-Critical Engineers</title><link>https://speytech.com/insights/hello-world-fails-safety-critical/</link><guid isPermaLink="true">https://speytech.com/insights/hello-world-fails-safety-critical/</guid><description>Traditional C tutorials teach habits that certification auditors reject</description><pubDate>Mon, 26 Jan 2026 18:30:00 GMT</pubDate><category>Safety-Critical Programming</category><author>william@fstopify.com (William Murray)</author></item><item><title>C From Scratch: A New Approach to Learning Safety-Critical C</title><link>https://speytech.com/insights/c-from-scratch-book-launch/</link><guid isPermaLink="true">https://speytech.com/insights/c-from-scratch-book-launch/</guid><description>Why proving code correct before writing it changes everything</description><pubDate>Sat, 24 Jan 2026 22:31:00 GMT</pubDate><category>Announcements</category><author>william@fstopify.com (William Murray)</author></item><item><title>Testing ML Systems: Beyond Unit Tests and Accuracy Metrics</title><link>https://speytech.com/ai-architecture/ml-testing-systems/</link><guid isPermaLink="true">https://speytech.com/ai-architecture/ml-testing-systems/</guid><description>A practical testing strategy for production machine learning</description><pubDate>Sat, 24 Jan 2026 00:30:00 GMT</pubDate><category>AI Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>Cost Engineering for ML Infrastructure: What Actually Matters</title><link>https://speytech.com/ai-architecture/ml-cost-engineering/</link><guid isPermaLink="true">https://speytech.com/ai-architecture/ml-cost-engineering/</guid><description>Where the money goes and what to optimise first</description><pubDate>Sat, 24 Jan 2026 00:05:00 GMT</pubDate><category>AI Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>State Management in ML Services: Beyond Stateless Inference</title><link>https://speytech.com/ai-architecture/ml-state-management/</link><guid isPermaLink="true">https://speytech.com/ai-architecture/ml-state-management/</guid><description>Architectural patterns for ML systems that need to remember</description><pubDate>Fri, 23 Jan 2026 22:00:00 GMT</pubDate><category>AI Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>Graceful Degradation in ML Systems: When Your Model Can&apos;t Answer</title><link>https://speytech.com/ai-architecture/ml-graceful-degradation/</link><guid isPermaLink="true">https://speytech.com/ai-architecture/ml-graceful-degradation/</guid><description>Fallback strategies for production inference that fails gracefully instead of failing loudly</description><pubDate>Fri, 23 Jan 2026 21:14:00 GMT</pubDate><category>AI Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>The Observability Blind Spot: What ML Metrics Don&apos;t Tell You</title><link>https://speytech.com/ai-architecture/ml-observability-blind-spot/</link><guid isPermaLink="true">https://speytech.com/ai-architecture/ml-observability-blind-spot/</guid><description>Why accuracy looks fine while your production system burns</description><pubDate>Fri, 23 Jan 2026 18:00:00 GMT</pubDate><category>AI Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>Fixed-Point Fundamentals</title><link>https://speytech.com/open-source/fixed-point-fundamentals/</link><guid isPermaLink="true">https://speytech.com/open-source/fixed-point-fundamentals/</guid><description>Learn fixed-point arithmetic from first principles — because &apos;close enough&apos; isn&apos;t deterministic</description><pubDate>Thu, 22 Jan 2026 20:00:00 GMT</pubDate><category>Open Source</category><category>Education</category><author>william@fstopify.com (William Murray)</author></item><item><title>certifiable-bench</title><link>https://speytech.com/open-source/certifiable-bench/</link><guid isPermaLink="true">https://speytech.com/open-source/certifiable-bench/</guid><description>Performance benchmarking for deterministic ML — because &apos;fast&apos; means nothing if you can&apos;t prove it&apos;s correct</description><pubDate>Wed, 21 Jan 2026 22:00:00 GMT</pubDate><category>Open Source</category><category>Deterministic Computing</category><author>william@fstopify.com (William Murray)</author></item><item><title>Stochastic Rounding Without the Stochastic</title><link>https://speytech.com/insights/stochastic-rounding-deterministic/</link><guid isPermaLink="true">https://speytech.com/insights/stochastic-rounding-deterministic/</guid><description>How PRNG-controlled rounding can provide regularisation benefits deterministically</description><pubDate>Tue, 20 Jan 2026 22:00:00 GMT</pubDate><category>Deterministic Computing</category><author>william@fstopify.com (William Murray)</author></item><item><title>Cross-Platform Bit-Identity: From Theory to 7 Matching Hashes</title><link>https://speytech.com/insights/cross-platform-bit-identity/</link><guid isPermaLink="true">https://speytech.com/insights/cross-platform-bit-identity/</guid><description>The practical journey of verifying deterministic ML across platforms</description><pubDate>Tue, 20 Jan 2026 20:45:00 GMT</pubDate><category>Deterministic Computing</category><author>william@fstopify.com (William Murray)</author></item><item><title>The Feistel Shuffle: Deterministic Data Ordering Without Randomness</title><link>https://speytech.com/insights/feistel-shuffle-deterministic/</link><guid isPermaLink="true">https://speytech.com/insights/feistel-shuffle-deterministic/</guid><description>How cycle-walking Feistel networks can provide reproducible shuffling for ML training</description><pubDate>Tue, 20 Jan 2026 20:24:00 GMT</pubDate><category>Deterministic Computing</category><author>william@fstopify.com (William Murray)</author></item><item><title>Merkle Chains for ML Audit Trails</title><link>https://speytech.com/insights/merkle-chains-ml-audit/</link><guid isPermaLink="true">https://speytech.com/insights/merkle-chains-ml-audit/</guid><description>How cryptographic hash chains can make every training step verifiable</description><pubDate>Tue, 20 Jan 2026 19:30:00 GMT</pubDate><category>Deterministic Computing</category><author>william@fstopify.com (William Murray)</author></item><item><title>Round-to-Nearest-Even: The Rounding Mode That Makes Determinism Possible</title><link>https://speytech.com/insights/round-to-nearest-even/</link><guid isPermaLink="true">https://speytech.com/insights/round-to-nearest-even/</guid><description>Why banker&apos;s rounding matters for bit-identical machine learning</description><pubDate>Tue, 20 Jan 2026 19:00:00 GMT</pubDate><category>Deterministic Computing</category><author>william@fstopify.com (William Murray)</author></item><item><title>The Certifiable-* Ecosystem: Eight Projects, One Deterministic ML Pipeline</title><link>https://speytech.com/ai-architecture/certifiable-ecosystem/</link><guid isPermaLink="true">https://speytech.com/ai-architecture/certifiable-ecosystem/</guid><description>From training data to deployed inference — bit-identical, auditable, certifiable</description><pubDate>Mon, 19 Jan 2026 23:00:00 GMT</pubDate><category>AI Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>certifiable-harness</title><link>https://speytech.com/open-source/certifiable-harness/</link><guid isPermaLink="true">https://speytech.com/open-source/certifiable-harness/</guid><description>End-to-end test harness for deterministic ML — because &apos;it works on my machine&apos; isn&apos;t certifiable</description><pubDate>Mon, 19 Jan 2026 22:00:00 GMT</pubDate><category>Open Source</category><category>Deterministic ML</category><author>william@fstopify.com (William Murray)</author></item><item><title>certifiable-verify</title><link>https://speytech.com/open-source/certifiable-verify/</link><guid isPermaLink="true">https://speytech.com/open-source/certifiable-verify/</guid><description>Pipeline verification for the certifiable-* ecosystem — because &apos;we checked it manually&apos; isn&apos;t certifiable</description><pubDate>Mon, 19 Jan 2026 21:00:00 GMT</pubDate><category>Open Source</category><category>Deterministic ML</category><author>william@fstopify.com (William Murray)</author></item><item><title>certifiable-monitor</title><link>https://speytech.com/open-source/certifiable-monitor/</link><guid isPermaLink="true">https://speytech.com/open-source/certifiable-monitor/</guid><description>Deterministic runtime monitoring — because &apos;the model drifted&apos; isn&apos;t certifiable</description><pubDate>Mon, 19 Jan 2026 17:00:00 GMT</pubDate><category>Open Source</category><category>Open Source</category><author>william@fstopify.com (William Murray)</author></item><item><title>A Complete Deterministic ML Pipeline for Safety-Critical Systems</title><link>https://speytech.com/ai-architecture/deterministic-ml-pipeline/</link><guid isPermaLink="true">https://speytech.com/ai-architecture/deterministic-ml-pipeline/</guid><description>From training data to deployed inference — bit-identical, auditable, certifiable</description><pubDate>Mon, 19 Jan 2026 00:15:00 GMT</pubDate><category>AI Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>Certifiable Deploy</title><link>https://speytech.com/open-source/certifiable-deploy/</link><guid isPermaLink="true">https://speytech.com/open-source/certifiable-deploy/</guid><description>Deterministic model packaging and cryptographic attestation — because &apos;trust me, it&apos;s the right model&apos; isn&apos;t certifiable</description><pubDate>Sun, 18 Jan 2026 22:00:00 GMT</pubDate><category>Open Source</category><category>Deterministic ML</category><author>william@fstopify.com (William Murray)</author></item><item><title>Certifiable Quant</title><link>https://speytech.com/open-source/certifiable-quant/</link><guid isPermaLink="true">https://speytech.com/open-source/certifiable-quant/</guid><description>Deterministic model quantization with formal error certificates for safety-critical ML</description><pubDate>Sun, 18 Jan 2026 21:00:00 GMT</pubDate><category>Open Source</category><category>Deterministic ML</category><author>william@fstopify.com (William Murray)</author></item><item><title>C-From-Scratch</title><link>https://speytech.com/open-source/c-from-scratch/</link><guid isPermaLink="true">https://speytech.com/open-source/c-from-scratch/</guid><description>Learn to build safety-critical systems in C — mathematical rigour, not &apos;Hello World&apos;</description><pubDate>Sat, 17 Jan 2026 09:00:00 GMT</pubDate><category>Open Source</category><category>Education</category><author>william@fstopify.com (William Murray)</author></item><item><title>Certifiable Data</title><link>https://speytech.com/open-source/certifiable-data/</link><guid isPermaLink="true">https://speytech.com/open-source/certifiable-data/</guid><description>Deterministic data pipelines for safety-critical ML — because &apos;we shuffled the data&apos; isn&apos;t reproducible</description><pubDate>Fri, 16 Jan 2026 10:00:00 GMT</pubDate><category>Open Source</category><category>Deterministic ML</category><author>william@fstopify.com (William Murray)</author></item><item><title>C-Sentinel</title><link>https://speytech.com/open-source/c-sentinel/</link><guid isPermaLink="true">https://speytech.com/open-source/c-sentinel/</guid><description>Semantic observability for UNIX systems — lightweight system probing with explainable risk scoring</description><pubDate>Fri, 16 Jan 2026 09:10:00 GMT</pubDate><category>Open Source</category><category>Security Monitoring</category><author>william@fstopify.com (William Murray)</author></item><item><title>WCET Analysis for Neural Network Inference</title><link>https://speytech.com/ai-architecture/wcet-neural-network-inference/</link><guid isPermaLink="true">https://speytech.com/ai-architecture/wcet-neural-network-inference/</guid><description>How to prove worst-case execution time for convolution, matrix multiply, and pooling operations</description><pubDate>Thu, 15 Jan 2026 22:31:00 GMT</pubDate><category>AI Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>Fixed-Point Neural Networks: The Math Behind Q16.16</title><link>https://speytech.com/insights/fixed-point-neural-networks/</link><guid isPermaLink="true">https://speytech.com/insights/fixed-point-neural-networks/</guid><description>How integer arithmetic can enable deterministic AI inference for safety-critical systems</description><pubDate>Thu, 15 Jan 2026 21:30:00 GMT</pubDate><category>Safety-Critical AI</category><author>william@fstopify.com (William Murray)</author></item><item><title>Why TensorFlow Lite Faces Challenges in DO-178C Certification</title><link>https://speytech.com/ai-architecture/tflite-do178c-challenges/</link><guid isPermaLink="true">https://speytech.com/ai-architecture/tflite-do178c-challenges/</guid><description>Understanding the architectural properties that complicate aerospace certification for mobile inference frameworks</description><pubDate>Thu, 15 Jan 2026 20:45:00 GMT</pubDate><category>AI Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>Bit-Perfect Reproducibility: Why It Matters and How to Prove It</title><link>https://speytech.com/insights/bit-perfect-reproducibility/</link><guid isPermaLink="true">https://speytech.com/insights/bit-perfect-reproducibility/</guid><description>What deterministic execution actually means and how to verify it across platforms</description><pubDate>Thu, 15 Jan 2026 20:15:00 GMT</pubDate><category>Safety-Critical AI</category><author>william@fstopify.com (William Murray)</author></item><item><title>The Real Cost of Dynamic Memory in Safety-Critical Systems</title><link>https://speytech.com/insights/dynamic-memory-safety-critical/</link><guid isPermaLink="true">https://speytech.com/insights/dynamic-memory-safety-critical/</guid><description>Why malloc is problematic for certification and how static allocation can simplify verification</description><pubDate>Thu, 15 Jan 2026 19:20:00 GMT</pubDate><category>Safety-Critical AI</category><author>william@fstopify.com (William Murray)</author></item><item><title>Certifiable Inference</title><link>https://speytech.com/open-source/certifiable-inference/</link><guid isPermaLink="true">https://speytech.com/open-source/certifiable-inference/</guid><description>Deterministic, bit-perfect neural network inference for safety-critical systems</description><pubDate>Thu, 15 Jan 2026 10:00:00 GMT</pubDate><category>Open Source</category><category>Deterministic ML</category><author>william@fstopify.com (William Murray)</author></item><item><title>Why Floating Point Is Dangerous: The Case for Deterministic AI in C</title><link>https://speytech.com/ai-architecture/floating-point-danger/</link><guid isPermaLink="true">https://speytech.com/ai-architecture/floating-point-danger/</guid><description>When &apos;mostly reproducible&apos; isn&apos;t good enough for systems that matter</description><pubDate>Wed, 14 Jan 2026 20:00:00 GMT</pubDate><category>AI Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>Certifiable Training</title><link>https://speytech.com/open-source/certifiable-training/</link><guid isPermaLink="true">https://speytech.com/open-source/certifiable-training/</guid><description>Deterministic ML training with Merkle audit trails — because &apos;we trained it&apos; isn&apos;t certifiable</description><pubDate>Wed, 14 Jan 2026 10:00:00 GMT</pubDate><category>Open Source</category><category>Deterministic ML</category><author>william@fstopify.com (William Murray)</author></item><item><title>Debugging Model Behavior in Production</title><link>https://speytech.com/ai-architecture/debugging-model-behavior-production/</link><guid isPermaLink="true">https://speytech.com/ai-architecture/debugging-model-behavior-production/</guid><description>When the model works in staging but fails in prod, here&apos;s how to find out why</description><pubDate>Tue, 13 Jan 2026 21:40:00 GMT</pubDate><category>AI Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>When You Don&apos;t Need a Feature Store</title><link>https://speytech.com/ai-architecture/when-you-dont-need-feature-store/</link><guid isPermaLink="true">https://speytech.com/ai-architecture/when-you-dont-need-feature-store/</guid><description>Most teams solve a problem they don&apos;t have yet</description><pubDate>Tue, 13 Jan 2026 21:15:00 GMT</pubDate><category>AI Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>Model Serving Architecture Patterns</title><link>https://speytech.com/ai-architecture/ai-model-serving-patterns/</link><guid isPermaLink="true">https://speytech.com/ai-architecture/ai-model-serving-patterns/</guid><description>Understanding latency, throughput, and the trade-offs between them</description><pubDate>Tue, 13 Jan 2026 20:35:00 GMT</pubDate><category>AI Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>Production AI Systems: What 30 Years of UNIX Taught Me</title><link>https://speytech.com/ai-architecture/production-ai-unix-principles/</link><guid isPermaLink="true">https://speytech.com/ai-architecture/production-ai-unix-principles/</guid><description>The infrastructure principles that kept systems running still apply to ML</description><pubDate>Tue, 13 Jan 2026 19:51:00 GMT</pubDate><category>AI Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>The Observability Gap in ML Systems</title><link>https://speytech.com/ai-architecture/ml-observability-gap/</link><guid isPermaLink="true">https://speytech.com/ai-architecture/ml-observability-gap/</guid><description>Why your model serving cluster fails at 3AM and you can&apos;t figure out why</description><pubDate>Tue, 13 Jan 2026 19:15:00 GMT</pubDate><category>AI Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>From Proofs to Code: Mathematical Transcription in C</title><link>https://speytech.com/insights/mathematical-proofs-to-code/</link><guid isPermaLink="true">https://speytech.com/insights/mathematical-proofs-to-code/</guid><description>How mathematical contracts become deterministic implementations</description><pubDate>Mon, 12 Jan 2026 23:25:00 GMT</pubDate><category>Formal Methods</category><author>william@fstopify.com (William Murray)</author></item><item><title>The Hidden Cost of Non-Determinism</title><link>https://speytech.com/insights/nondeterminism-cost-debugging/</link><guid isPermaLink="true">https://speytech.com/insights/nondeterminism-cost-debugging/</guid><description>Understanding the financial impact of debugging race conditions and Heisenbugs</description><pubDate>Mon, 12 Jan 2026 19:50:00 GMT</pubDate><category>Business Case</category><author>william@fstopify.com (William Murray)</author></item><item><title>ISO 26262 and ASIL-D: The Role of Determinism</title><link>https://speytech.com/insights/iso-26262-asil-d-determinism/</link><guid isPermaLink="true">https://speytech.com/insights/iso-26262-asil-d-determinism/</guid><description>How deterministic architecture can support automotive functional safety certification</description><pubDate>Mon, 12 Jan 2026 19:20:00 GMT</pubDate><category>Certification</category><author>william@fstopify.com (William Murray)</author></item><item><title>The Hidden Cost of &apos;Consistent Environments&apos;: What Docker Actually Guarantees</title><link>https://speytech.com/insights/docker-hidden-cost/</link><guid isPermaLink="true">https://speytech.com/insights/docker-hidden-cost/</guid><description>Why shipping 450MB to avoid inconsistency may be solving the wrong problem</description><pubDate>Mon, 12 Jan 2026 18:30:00 GMT</pubDate><category>Systems Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>The Zero-Variance Problem: When Mathematics Meets Machines</title><link>https://speytech.com/insights/zero-variance-problem/</link><guid isPermaLink="true">https://speytech.com/insights/zero-variance-problem/</guid><description>Why perfect consistency creates impossible calculations—and how to handle it</description><pubDate>Sun, 11 Jan 2026 22:50:00 GMT</pubDate><category>Formal Methods</category><author>william@fstopify.com (William Murray)</author></item><item><title>Composition Without Compromise: Connecting FSMs</title><link>https://speytech.com/insights/fsm-composition/</link><guid isPermaLink="true">https://speytech.com/insights/fsm-composition/</guid><description>How verified components combine to create verified systems</description><pubDate>Sun, 11 Jan 2026 22:22:00 GMT</pubDate><category>Formal Methods</category><author>william@fstopify.com (William Murray)</author></item><item><title>Why Your ML Model Gives Different Results Every Tuesday</title><link>https://speytech.com/insights/ml-nondeterminism-problem/</link><guid isPermaLink="true">https://speytech.com/insights/ml-nondeterminism-problem/</guid><description>The hidden sources of non-determinism in machine learning, and why they matter more than you think</description><pubDate>Fri, 09 Jan 2026 21:20:00 GMT</pubDate><category>Deterministic Computing</category><author>william@fstopify.com (William Murray)</author></item><item><title>Closure, Totality, and the Algebra of Safe Systems</title><link>https://speytech.com/insights/closure-totality-algebra/</link><guid isPermaLink="true">https://speytech.com/insights/closure-totality-algebra/</guid><description>Why three mathematical properties can determine whether your system fails safe or fails silent</description><pubDate>Thu, 08 Jan 2026 21:20:00 GMT</pubDate><category>Formal Methods</category><author>william@fstopify.com (William Murray)</author></item><item><title>Heartbeats and State Machines: Proving Existence in Time</title><link>https://speytech.com/insights/heartbeats-state-machines/</link><guid isPermaLink="true">https://speytech.com/insights/heartbeats-state-machines/</guid><description>How deterministic finite state machines can model liveness detection with verifiable properties</description><pubDate>Thu, 08 Jan 2026 17:40:00 GMT</pubDate><category>Formal Methods</category><author>william@fstopify.com (William Murray)</author></item><item><title>IEC 62304 Class C: What Medical Device Software Actually Requires</title><link>https://speytech.com/insights/iec-62304-class-c/</link><guid isPermaLink="true">https://speytech.com/insights/iec-62304-class-c/</guid><description>How deterministic architecture can support the most stringent safety classification for medical device software</description><pubDate>Thu, 08 Jan 2026 17:15:00 GMT</pubDate><category>Medical Devices</category><author>william@fstopify.com (William Murray)</author></item><item><title>Statistics as State Transitions</title><link>https://speytech.com/insights/statistics-as-state-transitions/</link><guid isPermaLink="true">https://speytech.com/insights/statistics-as-state-transitions/</guid><description>How finite state machines can provide a deterministic foundation for anomaly detection</description><pubDate>Wed, 07 Jan 2026 21:00:00 GMT</pubDate><category>Formal Methods</category><author>william@fstopify.com (William Murray)</author></item><item><title>Why EMA Beats SMA for Safety-Critical Monitoring</title><link>https://speytech.com/insights/ema-vs-sma-monitoring/</link><guid isPermaLink="true">https://speytech.com/insights/ema-vs-sma-monitoring/</guid><description>How exponential moving averages provide the mathematical properties that safety-critical systems require</description><pubDate>Wed, 07 Jan 2026 00:00:00 GMT</pubDate><category>Systems Architecture</category><author>william@fstopify.com (William Murray)</author></item><item><title>Semantic Security Monitoring in Deterministic Systems</title><link>https://speytech.com/insights/semantic-security-monitoring/</link><guid isPermaLink="true">https://speytech.com/insights/semantic-security-monitoring/</guid><description>How structured observability can support incident investigation and regulatory compliance</description><pubDate>Wed, 07 Jan 2026 00:00:00 GMT</pubDate><category>Safety Engineering</category><author>william@fstopify.com (William Murray)</author></item><item><title>MDCP vs. Conventional RTOS</title><link>https://speytech.com/insights/mdcp-vs-conventional-rtos/</link><guid isPermaLink="true">https://speytech.com/insights/mdcp-vs-conventional-rtos/</guid><description>Side-by-side comparison of MDCP deterministic platform against VxWorks, QNX, FreeRTOS, and conventional real-time operating systems</description><pubDate>Mon, 05 Jan 2026 00:00:00 GMT</pubDate><category>Technical Comparison</category><author>william@fstopify.com (William Murray)</author></item><item><title>NVIDIA&apos;s ASIL-B Linux Kernel: The Path to ASIL-D and Deterministic Execution</title><link>https://speytech.com/insights/nvidia-asil-determinism/</link><guid isPermaLink="true">https://speytech.com/insights/nvidia-asil-determinism/</guid><description>NVIDIA&apos;s achievement is impressive — and it clarifies the architectural considerations for Level 4/5 autonomy</description><pubDate>Wed, 24 Dec 2025 00:00:00 GMT</pubDate><category>Autonomous Systems</category><author>william@fstopify.com (William Murray)</author></item><item><title>Reproducibility and the Economics of Debugging</title><link>https://speytech.com/insights/debugging-economics/</link><guid isPermaLink="true">https://speytech.com/insights/debugging-economics/</guid><description>How non-deterministic execution can amplify engineering cost and schedule risk</description><pubDate>Tue, 23 Dec 2025 00:00:00 GMT</pubDate><category>Engineering Economics</category><author>william@fstopify.com (William Murray)</author></item><item><title>The Mathematics of Deterministic Scheduling</title><link>https://speytech.com/insights/deterministic-scheduling/</link><guid isPermaLink="true">https://speytech.com/insights/deterministic-scheduling/</guid><description>How tick-based execution addresses non-determinism at the fundamental level of computation</description><pubDate>Tue, 23 Dec 2025 00:00:00 GMT</pubDate><category>Formal Methods</category><author>william@fstopify.com (William Murray)</author></item><item><title>Cryptographic Execution Tracing and Evidentiary Integrity</title><link>https://speytech.com/insights/cryptographic-proof-execution/</link><guid isPermaLink="true">https://speytech.com/insights/cryptographic-proof-execution/</guid><description>How deterministic systems with cryptographic sealing can strengthen post-incident analysis</description><pubDate>Tue, 23 Dec 2025 00:00:00 GMT</pubDate><category>Deterministic Computing</category><author>william@fstopify.com (William Murray)</author></item><item><title>DO-178C Level A Certification: How Deterministic Execution Can Streamline Certification Effort</title><link>https://speytech.com/insights/do178c-certification/</link><guid isPermaLink="true">https://speytech.com/insights/do178c-certification/</guid><description>How tick-based architecture and minimal code footprint can reduce certification effort and timelines in aerospace programs</description><pubDate>Tue, 23 Dec 2025 00:00:00 GMT</pubDate><category>Safety Certification</category><author>william@fstopify.com (William Murray)</author></item><item><title>Reproducibility and Post-Incident Analysis in Implantable Cardiac Devices</title><link>https://speytech.com/insights/implantable-device-litigation/</link><guid isPermaLink="true">https://speytech.com/insights/implantable-device-litigation/</guid><description>How execution traceability can influence investigation, liability, and regulatory outcomes</description><pubDate>Tue, 23 Dec 2025 00:00:00 GMT</pubDate><category>Medical Devices</category><author>william@fstopify.com (William Murray)</author></item><item><title>The MycoEco Kernel: Lock-Free Determinism at Scale</title><link>https://speytech.com/insights/mycoeco-kernel/</link><guid isPermaLink="true">https://speytech.com/insights/mycoeco-kernel/</guid><description>How event-based architecture enables deterministic multi-core execution without locks</description><pubDate>Tue, 23 Dec 2025 00:00:00 GMT</pubDate><category>Systems Architecture</category><author>william@fstopify.com (William Murray)</author></item></channel></rss>