Of the ten postings, Twitter's is the most revealing — not because it's about a flashy product, but because it admits something most big-tech postings hide: they're rebuilding capacity management from scratch, and the bottleneck has become organizational, not technical.

The stack signal. The posting is deliberately stack-agnostic ("frontend/backend/fullstack, data science, and engineering management"), which tells you this is a greenfield internal-tools team, not a feature group slotting into an existing codebase. They're hiring across the full vertical at once — a pattern that usually means leadership has secured headcount but hasn't yet committed to architecture. Expect Scala/JVM on the backend (Twitter's historical lean), with Python for the forecasting/data-pipeline work and React on the dashboards.

What it reveals about the company. Three numbers do the heavy lifting: >100k physical servers, "expanding footprint in public cloud," and "supporting partners in finance and supply." Translation:

• Twitter is still running a massive on-prem fleet but is no longer pretending the cloud migration is optional.
• The CFO's office is now a first-class stakeholder in infra decisions — capacity is being treated as a financial planning problem, not just an SRE problem.
• They previously had a capacity system (this is described as a "new iteration"), and it failed at the human-process layer — the new pitch is "stronger product focus on enabling engineers." That's code for: the old system was ignored or bypassed.

Skills and trends highlighted. This posting sits squarely in the FinOps wave — the discipline of treating cloud and infra spend as a forecasting/optimization problem with proper tooling. The specific skill cocktail (demand forecasting + supply tracking + allocation prioritization + utilization reporting) is essentially supply-chain management applied to compute. Expect to see more of these teams across every hyperscaler customer in 2020-2021 as cloud bills hit board-level visibility.

Green flags: Clear problem statement, named internal customers (finance, supply), acknowledgment that the prior attempt under-delivered, and a remote option from a company historically reluctant to offer one.

Red flags: Hiring an entire org — ICs, data scientists, and managers — simultaneously is risky; the team will spend its first six months arguing about scope. "Internal tooling for finance partners" is also a notoriously thankless surface area where success is invisible and failure shows up in someone's quarterly earnings call.