Advanced terms
Client-Side Template Injection (CSTI)
Attacker-controlled input rendered as a template expression by a client-side framework (AngularJS, Vue, Handlebars), executing JavaScript in the victim's browser.
CWE-1336 OWASP A3:2021
1w ago
security advanced
7.5
Constitutional AI (CAI)
Anthropic's training methodology where models critique and revise their own outputs against a set of written principles, reducing reliance on human labellers for alignment.
1w ago
ai_ml advanced
CSS Injection & Data Exfiltration via Stylesheets
Attacker-controlled CSS injected into a page or stylesheet that exfiltrates data via attribute selectors and `url()` callbacks, defaces UI, or enables phishing — all without a single line of JavaScript.
CWE-79 OWASP A3:2021
1w ago
security advanced
6.5
DOM Clobbering
Attack where injected HTML elements with controlled `id` or `name` attributes overwrite JavaScript globals or document properties, weaponising script-less HTML injection into code execution.
CWE-79 OWASP A3:2021
1w ago
security advanced
6.5
Mixture of Experts (MoE)
Neural network architecture where a gating network routes each token to a small subset of specialist 'expert' sub-networks, enabling huge total parameter counts at moderate per-token compute cost.
1w ago
ai_ml advanced
RLHF — Reinforcement Learning from Human Feedback
Post-training method where human preference rankings train a reward model that fine-tunes an LLM via reinforcement learning, aligning outputs with human preferences.
1w ago
ai_ml advanced
cgroups — Linux Control Groups
A Linux kernel feature that groups processes and meters or limits their access to CPU, memory, I/O, network, and other resources — the core primitive that makes Docker, systemd, and Kubernetes resource limits possible.
3w ago
linux advanced
Diffusion Models
A class of generative models that learn to reverse a gradual noising process — starting from pure noise and iteratively denoising into coherent images, audio or video; the core technique behind Stable Diffusion, Midjourney and DALL·E 3.
3w ago
ai_ml advanced
Variance describes how subtype relationships on a type's components affect subtype relationships on the whole type — covariant types follow the direction, contravariant types reverse it.
4w ago
typescript advanced
Module augmentation lets you extend third-party or global type definitions without modifying their source — adding properties to existing interfaces, modules, or the global scope.
4w ago
typescript advanced
Types that reference themselves to describe arbitrarily nested structures — trees, nested menus, JSON, linked lists — without requiring any escape hatch.
4w ago
typescript advanced
Insecure Deserialization
Deserializing attacker-controlled data can trigger arbitrary object construction and method calls — PHP's unserialize() with untrusted input enables remote code execution via gadget chains in the loaded class graph.
CWE-502 OWASP A8:2021
4w ago
security advanced
Timing Attacks
Side-channel attacks that infer secret values by measuring how long an operation takes — a string comparison that short-circuits on the first mismatch leaks information about the secret one character at a time.
CWE-208
4w ago
security advanced
Memory-Mapped Files
A file mapped directly into a process's virtual address space — reads and writes go through the OS page cache rather than read()/write() syscalls, enabling fast access to large files and shared memory between processes.
1mo ago
linux advanced
Mutex vs Semaphore PHP 7.0+
A mutex allows only one thread to access a resource at a time — a semaphore controls access to a pool of N identical resources.
CWE-362
1mo ago
concurrency advanced
Optimistic Locking
A concurrency pattern using a version column to detect conflicting concurrent writes — no row locks held between read and write.
1mo ago
database advanced
SELECT FOR UPDATE
A locking read that acquires exclusive row locks — preventing other transactions from modifying selected rows until commit.
1mo ago
database advanced
WebAssembly (Wasm)
A binary instruction format that runs at near-native speed in the browser and on servers — enabling C, Rust, and Go code to run alongside JavaScript without plugins.
1mo ago
frontend advanced
The research and engineering discipline of ensuring AI systems pursue goals that are consistent with human values, intentions, and safety — not just stated objectives.
1mo ago
ai_ml advanced
An adversarial technique where malicious instructions are injected into an LLM's context window — via user input, retrieved documents, or tool results — to hijack the model's behaviour.
1mo ago
ai_ml advanced