How to Choose What to Practice

What you will be able to do#

• Choose practice from a concrete prerequisite, reasoning, implementation, or test gap.
• Use Foundry 75, foundations, extensions, and role plans for their distinct purposes.
• Interpret public, browser-local, account-backed, and company-association states without overstating what they prove.

Begin with a small baseline check#

Choose two or three representative tasks, use the language you expect to interview in, and record where the attempt first became uncertain. Useful evidence names the obstacle: an unclear contract, missing prerequisite, unjustified pattern choice, broken invariant, slow implementation, incomplete tests, or incorrect complexity.

A solved count alone cannot tell you whether the reasoning was independent, familiar, hinted, or reconstructed after seeing a solution.

Choose one familiar language#

Open the language comparison only if syntax or standard-library recall is slowing down correct reasoning. Prefer the language in which you can implement and explain common collections reliably, provided the employer accepts it.

Use the Python or Java reference for targeted review. Language review supports the algorithm session; it is not a requirement to study every available language.

Select by prerequisite or weak skill#

Use a foundation when the failure is conceptual: complexity, recursion, correctness, or representation. Use a topic or pattern lesson when you need to derive a known technique. Use a Foundry 75 question when you need a bounded canonical rehearsal. Use an extension only after its prerequisites are stable and the target role or format makes it useful.

When recognition is the gap, hide labels and state prior exposure before the attempt. When implementation is the gap, keep the technique known and measure whether you can produce correct code and tests.

Choose a plan after the baseline#

The roadmap shows dependencies; a time-boxed plan allocates sessions. Pick the role and time window closest to your confirmed process, then adjust for the evidence you already have. Do not treat a role label as proof that every advanced topic is required.

Use Practice, Roadmap, and Review as one loop#

Practice opens the public question catalog and lets you declare mode and prior exposure. Roadmap helps you choose prerequisites and sequence. Review returns to attempts that need reconstruction or a changed-context variant. Foundry 75 remains the finite canonical set; foundations teach missing reasoning, and extension exercises add transfer without inflating that 75.

1. Practice: attempt one bounded task under declared conditions.
2. Roadmap: repair the prerequisite that caused the first failure.
3. Review: reconstruct the reasoning and attempt a related variant separately.

Decide the next session from evidence#

Choose the smallest session that tests the failed reasoning. If you could not define prefix state, explain it before writing code. If you recognized a pattern but broke an index boundary, implement a focused variant. If both were stable, move to an unlabeled transfer task.

End with one explicit retrieval question for the next session rather than a vague instruction to practice more.

Know what progress states mean#

A status records activity, not mastery. Prior exposure and assistance change what an attempt can support. Anonymous visitors can read public lessons and use available browser-local practice; signed-in users may save private attempts and notes when account services are available.

When account services are unavailable, the product must say that account progress is unavailable. It must not claim a save, erase the public lesson, or reinterpret an unavailable account as signed out.

Treat company associations as scoped evidence#

The current DSA company browser labels its associations as demonstration data and does not show frequency or last-seen claims. A future verified association would establish only the claim and scope recorded by its source; it would not automatically become an interview-frequency estimate.

Choose core preparation from your gap and the employer's confirmed process, not from an unsupported company tag.

Worked example#

Apply the idea#

A learner solves familiar array tasks but fails unseen prefix-count problems. Select three sessions using the shipped Practice, Roadmap, and Review model.

Choose the next session#

Choose one question that tests the named gap; use the Roadmap only if a prerequisite must be repaired first.

Source and review record#

This original Engineering Foundry guide was checked by Codex against the recorded research scope on 2026-09-05. It has not received independent human editorial review.

Evidence IDs: SRC-DSA-TECH-HANDBOOK-PLAN SRC-DSA-ALGOMONSTER-ROADMAP