DigitableCourses
Portal settings
Русский
Show me around

Local and account-free. No account, no sign-up: whatever you type into the tools stays in this browser's localStorage and never reaches a server.

Our own counter records page opens and finishes: the address leaves, plus one tenth-of-the-text figure. No cookies, no outside trackers, your IP is not stored, Do Not Track is honoured. How to check that

The portal's own repository is not published, so we do not call it open source. What is open:

Digit flang Agent gate tools-core IT Tools Chat digitwm blender-mcp Hugging Face Licences

The portal lives on donations, paid consultations and requested write-ups, and on Workbench sales.

Support Order a write-up Workbench

There are no plans to make the courses paid.

Career

AI Engineer · LLMs and agents

Builds LLM applications with retrieval, agents, evaluation, safety and operations.

Roles and moves Self-assessment

Step 2 of 2 · AI Engineer · LLMs and agents

Which level inside the role?

Where people go from here: Backend developer, Data and machine learning, SRE and platform engineering, Architect, Team lead, tech lead, engineering manager

Not sure — open the whole role card Pick another role

  1. LLM application engineer
  2. AI Engineer
  3. Senior / Staff AI Engineer

LLM application engineer

Straight to the self-check

The focus moves from a prompt to a reproducible feature with checked output.

Do now

  • I compare the model with a solution without itBoth solutions run against the same examples and an explicit decision rule.
  • I define the output schema and check failuresInvalid output is handled explicitly before it reaches the product.
  • I build an evaluation setThe set includes normal cases, failures and unexpected input.

Where next

AI Engineer

Sideways out of the role

Skills, pitfalls and learning materials

What this step asks of you

  • I validate structured outputThe schema and refusal handling are tested against examples.
  • I compare quality on a fixed setInputs, expected results and model outputs are recorded.
  • I explain context and tokenization limitsI account for truncation and do not assume memory without storage.

Common mistakes

  • Treating fluent output as correct: errors reach the user.
  • Changing a prompt without rerunning checks: a fix breaks other examples.

What to learn

  1. LLM application engineer
  2. AI Engineer
  3. Senior / Staff AI Engineer

AI Engineer

Straight to the self-check

The scope expands from one answer to the reliability of the whole workflow.

Do now

  • I evaluate retrieval and generation separatelyI can locate retrieval misses separately from generation errors.
  • I constrain tool permissions and callsSensitive actions require explicit authorization outside the model output.
  • I measure request cost and latencyTraces show the model, context, calls, failures and cost.

Where next

Senior / Staff AI Engineer

Sideways out of the role

Skills, pitfalls and learning materials

What this step asks of you

  • I build retrieval with traceable sourcesAnswers reference documents and retrieval is evaluated independently.
  • I design a bounded agent loopLimits, failure handling and human handoff are explicit.
  • I test prompt injection and data boundariesUntrusted documents cannot acquire system instruction privileges.

Common mistakes

  • Treating retrieved content as instructions: injected text changes agent behavior.
  • Allowing unlimited retries: costs and side effects accumulate.

What to learn

  1. LLM application engineer
  2. AI Engineer
  3. Senior / Staff AI Engineer

Senior / Staff AI Engineer

Straight to the self-check

The outcome becomes a system that lets other teams ship their own features safely.

Do now

  • I choose between retrieval, fine tuning and conventional codeThe choice follows evidence about quality, cost and operational constraints.
  • I establish shared release evaluationsTeams reproduce evaluations and detect regressions before deployment.
  • I plan graceful degradation and model changesProvider failure leaves the product in a controlled state.

Where next

Not another step — a fork

Sideways out of the role

Skills, pitfalls and learning materials

What this step asks of you

  • I design LLM system operationsObservability, budgets, rollback and failure behavior are specified.
  • I justify whether fine tuning is neededI compare with the base model and retrieval on the same tasks.
  • I organize reproducible system evaluationTeams version datasets and analyze failure categories.

Common mistakes

  • Building a platform before teams need it: maintenance outweighs useful outcomes.
  • Choosing a model from one benchmark: behavior on product data remains unknown.

What to learn

IT Tools source · GPL-3.0 Digitable Chat source · GPL-2.0-or-later Legal documents