What Does a Software Engineer At Amazon Actually Do?

Key Concepts

• Service-Oriented Architecture (SOA): A design approach where systems are composed of small, isolated microservices owned by specific teams.
• On-Call: A rotation-based responsibility for maintaining system health and responding to incidents.
• Severity Levels (Sev 1–5): A classification system for incidents based on impact, where Sev 1 is the most critical.
• COE (Correction of Error): A post-mortem document used to analyze root causes of failures and define preventative action items.
• Design Documents (HLD/LLD): Technical blueprints required before implementation; HLD (High-Level Design) covers system-wide architecture, while LLD (Low-Level Design) covers specific implementation details.
• OP1/OP2 Cycles: Annual planning phases for setting roadmaps and project priorities.
• Two-Way Door Decisions: Reversible decisions that allow for flexibility and iteration.

---

1. Daily Routine (Micro Level)

The day-to-day life of an Amazon software engineer revolves around the software development life cycle (SDLC) within a service-oriented architecture.

• Coding: Java is the primary language (95% of teams), though Rust is used in performance-critical AWS applications. Engineers are encouraged to use AI tools to increase velocity.
• Testing & Verification: Includes unit tests, integration tests, and occasional manual testing for UI-heavy features.
• Debugging: Ranges from simple logic fixes to complex race conditions between distributed systems.
• Documentation: Maintaining internal wikis is critical to prevent redundant cross-team inquiries.
• Code Reviews: A mandatory process requiring at least one reviewer. High-quality reviews involve "putting on the hat of the author" to identify edge cases and anti-patterns.
• Meetings: Regular 1-on-1s with managers focus on career growth and project status. As engineers advance to SD2, SD3, and Principal levels, the focus shifts from individual coding to scaling the organization and mentoring others.

2. On-Call Responsibilities

On-call rotations are typically one week long, with frequency determined by team size (ideally 6–8 people).

• Ticketing:
  • Sev 3–5: Standard priority tickets; handled during normal working hours.
  • Sev 2: Significant system issues affecting customers. Requires immediate attention and often triggers a COE.
  • Sev 1: Critical outages (e.g., retail site down). Triggers an "all-hands-on-deck" response, including executives and cross-team coordination.
• Operational Excellence: On-call engineers must respond to pages within 15 minutes to avoid escalation. The week concludes with a handover meeting to discuss incidents, alarms, and pending tasks.

3. Project Cycles

Projects follow a structured design-first methodology:

• Design Phase: Engineers partner with Product Managers to convert a Business Requirements Document (BRD) into a technical design.
• Design Characteristics: Systems must be scalable, efficient, fault-tolerant, backwards compatible, secure, and observable.
• The "YAGNI" Principle: Engineers are cautioned against over-engineering ("You Ain't Gonna Need It") and are encouraged to prioritize "two-way door" decisions that can be reversed if requirements change.
• Review Process: Design documents undergo multiple rounds of peer review until the team reaches a consensus.

4. Seasonal Work & Planning

• OP1 (August): High-level brainstorming and alignment on annual goals and "moonshot" ideas.
• OP2 (Q4): Granular, refined planning where specific project scopes, estimates, and dependencies are finalized.
• Interviews: A recurring time commitment where engineers participate in hiring loops, including prep work and debrief sessions.
• Mentorship: A vital component for career growth, used to influence others and learn from senior technical leaders.

5. Key Perspectives & Quotes

• On Accountability: Regarding COEs, the speaker advises: "Don't try and deflect or blame it on other people. Just say like, 'Yeah, I effed up... It's our problem, our fault, and we need to do an autopsy on this.'"
• On Growth: Promotion criteria at Amazon shift as you move up; it is no longer just about your individual output, but about "improving the overall organization and the people around you."
• On Stress: The speaker notes that the specific, loud ringtone for Amazon pages can cause anxiety even for experienced engineers, emphasizing the high-pressure nature of on-call duties.

Synthesis

The role of a software engineer at Amazon is defined by a rigorous, document-heavy culture that prioritizes operational excellence and scalability. While the day-to-day involves standard coding and debugging, the "Amazon way" is characterized by a deep commitment to service ownership, a structured approach to incident management (COE), and a long-term planning cycle (OP1/OP2) that ensures alignment across a massive, distributed organization. Success at higher levels is measured by one's ability to scale systems and mentor others rather than just writing code.