底层系统理解
Search documents
OpenAI Codex 负责人:懂底层是没被淘汰的唯一底牌,顶尖工程师的终极归宿是“代码审查员”
AI科技大本营· 2026-03-11 09:51
Core Insights - The article discusses the evolution of engineering roles in major tech companies, particularly focusing on the transition from traditional coding to AI-assisted development, highlighting the experiences of Michael Bolin, a distinguished engineer at Meta and now at OpenAI [2][6][33]. Group 1: Career Progression and Challenges - Many senior engineers struggle to advance due to a tendency to focus on creating perfect solutions from scratch rather than addressing existing systemic issues that are critical to the company's success [11][30]. - The transition from Principal Engineer (E8) to Distinguished Engineer (E9) is often fraught with challenges, as the skills required shift from pure coding to political acumen and the ability to drive cross-departmental collaboration [25][30]. - Bolin's experience illustrates that the ability to write clear technical documentation and strategic plans is crucial for career advancement in large organizations [11][31]. Group 2: Cultural Shift in Tech Companies - The culture at OpenAI is research-led, contrasting with the engineering-led culture at companies like Meta and Google, where engineers are the primary decision-makers [34][36]. - This cultural shift can be jarring for engineers accustomed to having significant control over product direction, as they find themselves in supportive roles focused on enabling researchers [37][34]. Group 3: AI's Impact on Engineering - Bolin reports that AI now generates 80% to 90% of his code, fundamentally changing the role of engineers from writers to reviewers, where they focus on assessing AI-generated outputs for quality and correctness [6][41]. - The emergence of AI tools like Codex demonstrates the potential for AI to automate routine coding tasks, allowing engineers to concentrate on higher-level system architecture and design [39][41]. Group 4: Recommendations for Engineers - Deep technical skills remain essential, as AI can produce flawed code, and engineers must understand underlying systems to effectively evaluate AI outputs [44][45]. - Engaging in practical challenges, such as Capture The Flag (CTF) competitions, can enhance understanding of low-level systems and improve problem-solving skills [49].