TrainTrack

https://uselesstechjunks.github.io/traintrack/index.html

Short notes site that organizes preparation for a large-scale model training role (research engineer / scientist focused on training, objectives, data, and evaluation rather than product shipping). It is an imperfect guide aimed at senior-level research engineering roles centered on training large models at scale.

What this covers

• Ideal roles: research engineer/scientist (LLMs/foundation models), sometimes applied scientist or senior ML engineer if the focus is on training.
• Core mission: design, train, evaluate, and improve models directly via objectives, data, and optimization systems; success measured by model-level metrics.
• Day-to-day: training-time interventions, model quality improvement, evaluation/diagnostics, and systems that exist to support research.
• Evaluation expectations: ability to improve existing models, explain metric movement, work at scale, and translate fuzzy goals into concrete training changes.
• Resume stance: emphasize training-scale work, objectives, data decisions, offline metrics, and experiment reasoning; de-emphasize product metrics and feature delivery.

How to view

• Open docs/index.html locally to browse the notes; markdown files live in docs/content/ and render through docs/view.html.
• Key files:
  • READLIST.md: curated sources by tier, numbered with “Why” notes.
  • QUESTIONS.md: question-first prompts aligned to each tier.
  • SYLLABUS.md: condensed prep map tying the sections together.