Date and Time: 2025-05-13 12:30p ET

Future of Reliability Career from a consulting devops, DevSecOps role similar to SRE. job prospects vs traditional engineering infrastructure roles?
 - More focus on MLOps to learn what works well and poorly. careerwise: five-year future was unpredictable
 - Ensure systems are running vs pipelines and infrastructure. Job prospects
 - Reliability might be different: detect and minimize damage from hallucinations
 - Scott Galway: security = boring but hard for AI enter soon due to regulations, like logistics, health, and security.
 - Skill set: prompts for making agents that monitor availability.
 - Fixing support contract bottlenecks creates perceived value (even if self-inflicted).

Math Required for the top 10% of Reliability Engineering
 - how much math knowledge do you need for infra, reliability, sre, devops to be best at it? Probability, calculus, discrete math, etc?
 - Math PhD says: owning outcomes makes you best. Know: probability, calculus, linear algebra (for optimization: gradient descent, differentiate a surface), combinatorics. Model or simulate or find closed-form solution. More tools in your toolbox makes it more likely you have a useful tool. Top engineers get a paycheck, but so do others. Wisdom is in knowing what the packages do and whether they exist and are suitable.
 - Carpentry, bar tending, grocery shop was good enough for work, but without explainability. Do the risk analysis. TLA+ for mathematical verification of software. https://leanpub.com/logic watch the author’s SRECon talks on youtube
 - non-math person suggests: do a survey of math concepts, but you don't need to be expert in implementing them. Ask AI to apply the concepts to the problem.
 - eg: computing the p99 of multiple systems with their own p99s. Distributions are not normal. Look into queuing theory.

Quantifying the value of reliability. Are SREs able to communicate the value of uptime or negative value of downtime?
 - do others value it when it's present, or only when it's absent? Do you have tools to measure or predict?
 - SLI can be set at levels that do not affect users or are unnoticeable. Need budgets or targets to quantify business impact.
 - Customer-specific values: revenue, dev productivity, infrastructure/people/training return on investment, customer trust (communicating problems clearly can minimize impacts of downtime)