CConnectedResearchersinconnectedresearchers.hashnode.dev·Aug 9 · 7 min readPhoenix Treats Refresh Drift as a Recoverable Runtime StateThe DDR5 attack works because it exposes progress, detects a missed refresh, and repairs phase instead of demanding a perfect detector. Long-running systems fail differently from short routines. A det00
CConnectedResearchersinconnectedresearchers.hashnode.dev·Aug 8 · 8 min readDesigning LLM Assistants for High-Stakes Operations: Evidence Before VerdictsThe dominant product story for AI in a security operations centre is an autonomous copilot. It reads alerts, investigates telemetry, decides whether activity is malicious, and prepares or executes a r10
CConnectedResearchersinconnectedresearchers.hashnode.dev·Aug 7 · 6 min readTarget-Region Reconstruction as a Training Contract for Robot PoliciesA robot policy can recognize the instructed object and still move toward the wrong place. The gap sits between visual recognition and action generation. The target exists in the encoded scene, but the00
CConnectedResearchersinconnectedresearchers.hashnode.dev·Aug 6 · 7 min readPIPEMORPH: Treating Pipeline Parallel Training as a Runtime Control ProblemDistributed training systems often optimize the steady state. They choose a pipeline schedule, minimize bubbles on a healthy cluster, and assume the execution order will remain useful for the next ite00
CConnectedResearchersinconnectedresearchers.hashnode.dev·Aug 5 · 8 min readEncode Once, Query Any Point: The Interface Behind D4RTA developer-focused reading of the CVPR 2026 Best Paper: one global video representation, an independent five-index query, and task cost that follows requested output. Most dynamic video geometry syst00
