Jev System One vs Specialized Sovereign RAG. An Empirical Benchmark
Across 20 measured trials on Apple Silicon MPS, a cloud-hosted decision primitive averaged 457.19ms P50 (381.0ms upstream cloud floor) and consumed 453 tokens before retrieval began. A co-hosted sovereign classifier completed the same routing in 0.23ms with zero tokens.
Why Telling an LLM Not to Look Guarantees It Will
Vals AI's BioMysteryBench audit confirms what the attention mechanism already predicts: instructing a model to ignore data forces it to attend to that data. The fix is physical, not linguistic.
The 51.4% Problem: Why Post-Hoc Validation Fails Legal RAG Governance
Why post-hoc validation causes a 51.4% abstention rate in legal RAG governance, and how in-flight sentence-level NLI verification solves the hallucination tax.
Can You Build Confidential Legal AI Without a TEE? Empirical Lessons on PIR and Keyed Models
Two studies on fictional UK employment matters, October 2026, with the cloud provider treated as an attacker with full access. A model hidden behind a secret key was read in full: 100% of the secret content recovered from public statistics of its weights. Keeping client work on a laptop and fetching public law by private information retrieval sent 0 of 17 client secrets to the cloud, where a standard cloud pipeline sent 16 private tokens. With dates and amounts computed in code, 30 of 30 conclusions on 8 unseen matters were correct. Work that needs cloud compute on client data needs a hardware trusted execution environment, such as Prem Enclave.
In-Flight Rollback on SGLang: A Sentence-Level Check Cut Wrong Legal Answers From Memory by 83 to 86%
legal-rag-verifier checks each sentence while the model writes it, cuts a rejected one and rewrites it with the statutory text added. 240 UK statute questions answered from memory, 14 October 2026: answers holding a wrong sentence fell from 132 to 23 for Qwen 27B and from 198 to 28 for Qwen 2.5 7B. After a real error, the default retry was wrong again 71 times in 95, and a retry with the provision added at the cut 16 times. With the source already in the prompt, Qwen 27B had 4 such answers with the loop and 4 without, against 3 for Gemini 3.8 Flash and 8 for Claude Opus 4.6.
What Failed in My Own Controls: Seven Results From October 2026
Seven failures in my own legal AI controls, five measured against a rule set before the run and two found in review, each with its date, configuration and evidence: an auditor whose rejections were mostly wrong, a model-hiding scheme recovered 100%, routing that failed its pre-registered rule twice, two missed latency targets, a calculator fallback and a wrong statutory instrument on my own case study.
legal-rag-router: Deterministic Citation Routing for UK Legal RAG, Measured Against Gemini, Jev and Laya
An open-source router that binds a legal RAG query to the cited UK provision before vector search. Blind sealed run, 29 September 2026: 0 of 1,897 real citations bound to the wrong instrument. Gemini 3.8 Flash (high) routing on its own, 1 October 2026: 102 of 1,897.
The EU AI Act & Legal RAG: Why 'Approve' Buttons Fail Compliance Audits and How to Build Deterministic Human Oversight
From Article 14 Human Oversight to Entity Masking Engines: Architectural blueprints for engineering auditable, deterministic legal AI under strict EU/UK regulatory regimes.
Decoupling POSIX from Object Storage: The JuiceFS Architecture
Object storage is flat, but applications expect directories. A teardown of how JuiceFS splits the storage path, mitigates single-thread page fault traps in Redis, and implements 4MB chunk retrieval without local cache bloat.
When Safari Becomes the Silent Enemy
Spent days implementing video sessions. AWS failed. LiveKit crashed. Every major solution worked everywhere except iOS Safari. The problem wasn't the code. It was the platform's own interpretation of the standards. One architecture handled it; the others did not.
Context Engineering: The Hidden Art of Building with AI
Communication shapes better code than syntax ever could. On specification design, AI orchestration, and the emerging role that synthesizes systems rather than writes them.
Crafting a Digital Sanctum: A Contemplation on the Self-Hosted Home Server
A full stack developer's contemplation on building a self-hosted digital sanctum. Explore the philosophy and architecture of a bespoke home server for digital sovereignty, privacy, and on-demand power.
Code in JavaScript like a NASA Engineer
Explore coding guidelines inspired by NASA to write safe, reliable, and maintainable JavaScript/TypeScript code for real-world applications.
No entries found
Try a different search term.