Two strong reliability engineers left within a month of each other, both citing the same reason: no visible path forward that didn't mean becoming a manager.

Their exit interviews used almost identical language, which was the part that stung most. Both had been told, informally, that the only way to keep growing was to move into engineering management — a path neither of them wanted, and one that would have removed two of our strongest hands-on reliability engineers from the work they were actually great at.

The ladder we didn't have

We had a perfectly normal individual-contributor ladder on paper, up through staff and principal. What we didn't have was a version of it that felt real for reliability work specifically — every example of a "principal engineer" in our internal materials was a feature architect, and every promotion story employees had actually seen was the same shape. A ladder that exists on paper but has never been walked by anyone in your discipline doesn't feel like a path. It feels like a theoretical possibility.

reliability_ic_ladder (published with 3 real, named example career paths)
levels:
  - SRE II: owns reliability of one service area
  - Senior SRE: owns reliability across a domain, mentors on-call
  - Staff SRE: sets reliability strategy across multiple domains,
      blast-radius-scale prevention (see staff addendum)
  - Principal SRE: company-wide reliability strategy, represents
      reliability in exec-level tradeoff decisions
each_level_includes: >
  A real, named (with consent) example of someone at Helix currently
  at that level, with a 2-sentence description of what they actually
  do day to day — not a hypothetical description.
"I didn't need a promise. I needed to see one actual person, doing the job I wanted, five years further down the road than me."

What changed once the path was visible

We published the ladder with three real, consenting examples at Senior, Staff, and Principal level. Voluntary attrition on the reliability team dropped in the following year, and — a metric we hadn't anticipated — internal transfer requests into reliability roles from feature teams increased, because the path was now visibly real rather than a line item nobody had actually walked.

  • A career ladder that exists only on paper, with no real examples, doesn't function as a retention tool no matter how well it's worded.
  • Publish named, consenting examples at each level — an abstract description is much less persuasive than an actual person's day-to-day.
  • Losing engineers to "no path forward" is a ladder problem, not a compensation problem — check which one you actually have before reacting.