An open-source, evidence-based Codex coach for a beginner working toward an overseas remote DevOps role in 18 months.
This repository is not a list of links or a promise of employment. It combines a 78-week competency map, daily time-boxed practice, verified evidence, weekly adaptation, progressive technical English, portfolio gates, and a bridge into real operations experience.
The included learner profile assumes:
- a beginner in IT and approximately A2 English;
- 10–12 sustainable hours per week;
- AWS as the primary cloud track;
- a monthly learning and cloud budget of at most USD 20;
- overseas employee and long-term contractor roles as valid outcomes;
- junior operations, cloud support, and open-source contribution as experience bridges.
Remote DevOps roles commonly combine Linux, cloud infrastructure, Terraform, containers, CI/CD, observability, incident response, and strong written communication. Many current roles also ask for several years of production experience. The roadmap therefore targets job readiness and a credible opportunity pipeline; it does not guarantee a job.
- Generates or resumes one idempotent plan for a date.
- Preserves a single machine-readable progress state.
- Requires command output, code, tests, a runbook, or a demo before marking work complete.
- Schedules weak skills for review after two or seven days.
- Reduces, preserves, or increases the next week's load based on completion and mastery.
- Keeps at least 25% of learning time in English, including English project documentation.
- Assesses daily English through reading and writing; general speaking and pronunciation practice stays in Duolingo, while roadmap technical demos and mock interviews remain in scope.
- Keeps private application, company, income, and contact data outside Git.
- Never pushes daily progress until the learner explicitly says
完成并发布今日记录.
Requirements: Python 3.11 or later and Git.
python -m venv .venv
.venv\Scripts\Activate.ps1
python -m pip install -e ".[dev]"
python -m devops_coach validate
python -m devops_coach todayThen open the generated plan under plans/ or tell Codex:
开始今天学习
The durable coaching rules are in AGENTS.md. The learner-safe public settings are in config/learner.yml; copy config/learner.example.yml when adapting the project.
python -m devops_coach today [--date YYYY-MM-DD]
python -m devops_coach record --task ID --status done|partial|blocked --score 0..5 --minutes N --evidence PATH
python -m devops_coach review [--week YYYY-Www]
python -m devops_coach status
python -m devops_coach validate
Example evidence update:
python -m devops_coach record `
--task 2026-07-29-practice `
--status done `
--score 4 `
--minutes 25 `
--evidence evidence/week-01/wsl-environment.md| Day | English | Technical or project work | Review | Total |
|---|---|---|---|---|
| Monday–Friday | 20 min | 45 min | 10 min | 75 min |
| Saturday | 30 min + 15 min English README | 120 min project work | 15 min | 180 min |
| Sunday | 30 min | 45 min retrieval | 45 min review and planning | 120 min |
The weekday coach starts automatically at 09:00 and may include English reading and writing. General speaking and pronunciation practice is handled separately in Duolingo and is not tracked by this repository.
The project contains six 13-week phases: foundations, systems automation, containers and CI/CD, AWS and Terraform, Kubernetes and SRE, and a production capstone with the global job search. Every phase ends with a runnable public artifact, bilingual documentation, an English demo, and an evidence gate.
Before creating a potentially billable AWS resource, the coach must provide a cost estimate, require a budget alarm, receive explicit approval, and include a teardown command. Persistent EKS is out of scope for the USD 20 monthly budget; Kubernetes practice uses a local cluster by default.
The entire private/ directory is ignored. Public learning logs must not contain credentials, private keys, personal email addresses, employer data, internal hostnames, private repository names, or unredacted local paths. Run the scanner before publication:
python scripts/privacy_scan.pyThe weekday coach runs automatically at 09:00 and the weekend coach at 10:00 in Asia/Shanghai. Both may schedule English reading or writing, but they do not track general speaking, pronunciation, or recording practice. No API key is needed. See Codex automation setup.
Curriculum fixes, accessible beginner explanations, deterministic tests, and privacy-safe labs are welcome. Read CONTRIBUTING.md, SECURITY.md, and PRIVACY.md before opening a pull request.
MIT. See LICENSE.