테크 어휘
완전 정복

A neural network architecture based on self-attention mechanisms, foundational to modern large language models like GPT and BERT.

The process of further training a pre-trained model on a specific dataset to adapt it for a particular task or domain.

An optimization algorithm that iteratively adjusts model parameters to minimize the loss function during training.

A dense vector representation of data (words, images, etc.) that captures semantic relationships in a continuous space.

The process of using a trained machine learning model to make predictions on new, unseen data.

A phenomenon where an AI model generates plausible-sounding but factually incorrect or nonsensical output.

The practice of designing and optimizing input prompts to guide LLMs to produce more accurate and relevant outputs.

A technique combining retrieval of external knowledge with text generation, allowing LLMs to access up-to-date information.

When a model learns the training data too well, including noise and outliers, causing poor performance on new data.

The process of breaking down text into smaller units (tokens) that can be processed by a language model.

The time delay between submitting a request to an AI model and receiving its response.

The maximum amount of text (tokens) a language model can process in a single interaction.

A consensus mechanism where miners compete to solve complex mathematical puzzles to validate transactions and add new blocks.

Self-executing contracts with terms directly written into code, automatically enforcing and executing agreements without intermediaries.

A collection of funds locked in a smart contract used to facilitate decentralized trading and lending without traditional order books.

A schedule that releases tokens to team members or investors over time to align long-term incentives and prevent immediate selling.

An offline cryptocurrency storage method, not connected to the internet, providing enhanced security against remote hacks.

A fee paid to compensate network validators for the computing energy required to process and validate transactions on Ethereum.

A unique digital asset verified using blockchain technology, where each token is one-of-a-kind and cannot be replicated.

A protocol that ensures all nodes in a blockchain network agree on the same version of the distributed ledger.

A change to a blockchain's protocol, either soft (backward-compatible) or hard (non-backward-compatible), creating protocol divergence.

A technical document released by a blockchain project explaining its technology, use case, tokenomics, and development roadmap.

A protocol enabling assets or data to be transferred between two separate blockchain networks.

A series of 12-24 randomly generated words that act as the master key to recover a cryptocurrency wallet.

A decentralized exchange protocol using mathematical formulas to price assets instead of traditional buy/sell order books.

A strategy of moving crypto assets between protocols to maximize returns through transaction fees, interest, and governance tokens.

The process of locking cryptocurrency in a protocol to support network operations, earning rewards in return.

The total capital deposited into a DeFi protocol, used as a key metric to measure a protocol's adoption and health.

A temporary loss experienced by liquidity providers when the price ratio of deposited assets changes compared to when they were deposited.

A cryptocurrency that grants holders voting rights on protocol decisions, upgrades, and treasury management.

Continuous Integration/Continuous Deployment — automated workflows for building, testing, and deploying software changes rapidly.

Automated management of containerized applications across multiple hosts, handling deployment, scaling, and networking.

Managing and provisioning computing infrastructure using machine-readable configuration files rather than manual processes.

A deployment strategy maintaining two identical environments, switching traffic between them to enable zero-downtime releases.

Gradually rolling out new software to a small subset of users before deploying to the entire user base.

The ability to measure a system's current state based on the data it generates — logs, metrics, and traces.

A formal commitment between a service provider and client defining the expected performance and uptime standards.

The process of reverting an application or system to a previously stable version after a failed deployment or update.

An architectural pattern where applications are built as small, independent services that communicate via APIs.

A system's ability to handle increased workload by adding resources without compromising performance or reliability.

A design pattern where services communicate through events, enabling loose coupling and asynchronous processing.

A server that acts as an entry point for clients, routing requests to appropriate microservices and handling cross-cutting concerns.

A form of asynchronous service-to-service communication used to decouple applications and handle traffic spikes.

An application where all components are interconnected and deployed as a single unit, contrasted with microservices.

Command Query Responsibility Segregation — a pattern separating read and write operations into different models for performance optimization.

The implied cost of future rework caused by choosing a quick solution now instead of a better, more maintainable approach.

A systematic examination of source code intended to find and fix mistakes, improve quality, and share knowledge.

Restructuring existing code without changing its external behavior to improve readability and reduce complexity.

An agile technique where two developers work together at one workstation — one writes code while the other reviews in real-time.

A time-boxed iteration (typically 1-4 weeks) in Scrum where a team completes a set amount of work.

A method of debugging by explaining code line-by-line to an inanimate object, which forces articulation of assumptions.

Five design principles for object-oriented programming that make software designs more understandable, flexible, and maintainable.

A design pattern where an object receives its dependencies from external sources rather than creating them internally.

The amount of time a startup can continue operating at its current burn rate before running out of funding.

A fundamental shift in a startup's business model, product, or strategy in response to market feedback.

The degree to which a product satisfies strong market demand — the point where a business can efficiently scale growth.

Comprehensive appraisal of a business conducted by investors prior to completing a transaction or investment.

The rate at which a company spends its cash reserves before generating sufficient revenue to cover costs.

The percentage of customers who stop using a product or service over a given time period.