The engineering discipline of designing, building, and evaluating models that learn patterns from data to solve real-world problems.
Machine learning is the discipline of building models that learn patterns from data instead of following explicitly programmed rules. Grounded in statistics and linear algebra, it encompasses three learning paradigms: supervised, unsupervised, and reinforcement learning. You work across the full pipeline from data preparation to model deployment. Where general programming (programming) focuses on logic and structure, machine learning specializes in statistics-based modeling. It also differs from AI utilization (ai-utilization), which centers on using AI tools rather than building models.
You understand what machine learning is at a conceptual level and can distinguish supervised, unsupervised, and reinforcement learning. You spot ML at work in recommendation engines, spam filters, and voice assistants. You are learning how data feeds into model training and what makes a good dataset. This level precedes SFIA's defined ML scope, focusing on building the conceptual vocabulary that underpins all subsequent implementation stages.
What Comes Next
If you have checked off most of this list, you are ready to enter the Foundation Builder stage at SFIA's Assist level, where you implement and train basic ML models under guidance. Google MLCC(2024) places linear regression and classification as the first implementation milestones at this stage.
7단계 책임 수준(Follow→Assist→Apply→Enable→Ensure/Advise→Initiate/Influence→Set Strategy)으로 ML 역량을 정의. L2 안내하 기법 적용부터 L7 조직 전략 수립까지의 레벨 경계 설계의 1차 기준.
AI/ML 지식 영역의 역량 모델 — 신경망, 표현 학습, 강화학습, 생성 모델을 포함한 커리큘럼 체계. 수학/통계 요구사항의 단계적 심화가 L1-L4 체크리스트 설계 근거.
회귀→분류→신경망→임베딩→LLM→프로덕션의 모듈 시퀀스가 L2-L5 체크리스트 항목의 구체적 행동 지표. ML 공정성과 프로덕션 시스템 모듈이 L4-L5 윤리/배포 역량 근거.
수강생 480만 명 이상의 ML 입문 과정. 지도학습→비지도학습→강화학습 3단계 진행이 ML 학습 경로의 글로벌 표준. Andrew Ng의 교수법적 권위가 L1-L4 역량 범위 설정 근거.
ML 배포 워크플로우의 실무 사례 서베이. 데이터 관리, 모델 학습, 배포, 모니터링 각 단계의 도전 과제를 정리. L4-L6의 프로덕션 배포, MLOps, 조직 표준 수립 체크리스트 항목의 학술적 근거.
