< Notes />

MD Machine Learning Theory Intermediate

把失败归因给模型之前，先审 verifier——41% 的"模型错误"其实是判分 bug

跑一个自动判分的 benchmark，拿到 81% 通过率之后， 第一件事不是分析模型为什么错，是逐条看那 19 个失败案例。

2026-10-11 VIEW()

MD Security Engineering Advanced

24 of 24 SAST Findings Were False — The Real Bug Was Three Lines Away

Bandit flagged every mark_safe() call in the codebase. All 24 were false positives. The actual stor…

2026-07-30 VIEW()

MD Security Engineering Beginner

Audit the Deployed Set, Not the Lockfile

Scanning requirements.txt found nothing, because it could not even install. Scanning what the runni…

2026-07-30 VIEW()

MD Security Engineering Intermediate

Nginx add_header Does Not Inherit: How /media/ Lost Every Security Header

Security headers were configured once in the server block and applied to most of the site. Adding a…

2026-07-30 VIEW()

MD Security Engineering Intermediate

The Typo That Shipped DEBUG=True: Fail-Open Defaults and Env-Var Drift

A production Django site served traceback pages for months. No bad decision caused it — three envir…

2026-07-30 VIEW()

MD Zero-Knowledge Proofs Advanced

Tiered Verification and Red-Team Attacks on LLM Inference

A three-tier verification framework (SHA-256 commits → Fisher-selected proofs → full ZK) with 1000x…

2026-04-01 VIEW()

MD Zero-Knowledge Proofs Expert

Layerwise ZK Proofs for LLM Inference: Sumcheck and Constraint Reduction

How to cryptographically prove correct LLM inference without revealing model weights. Sumcheck redu…

2026-04-01 VIEW()

MD Machine Learning Theory Expert

FisherLD: When a Workshop-Accepted Method Turns Out to Be Provably Redundant

FisherLD proposed Fisher-guided layer-wise KD for LLM recommenders, got accepted at AAAI 2026 Works…

2026-04-01 VIEW()

MD Information Retrieval Advanced

GraphIndex: Incremental Multi-Modal KG Indexing Without LLMs

Three fused index modalities (structural, semantic, relational) with cross-modal incremental propag…

2026-04-01 VIEW()