DriveReady

Accenture Assessment #3 — Coding Round

60 minutes · 3 questions · one backend (Java) + one SQL (MySQL) + one frontend (HTML/CSS/JS) Platform: Mercer Mettl, Secure Exam Browser 1.6.8. Language: Java.

Start here

If you have… Read
5 minutes 91-CHEATSHEET.md
1 hour 03-ROUND-STRATEGY.md + solutions/java/JAVA-TRAPS.md §2
1 day 99-STUDY-PLAN.md → "If you have one day"
1 week or more 99-STUDY-PLAN.md
No HTML/JS at all frontend/learn/00-INDEX.md — the from-zero track, before anything frontend

The two highest-return actions, before any topic study:

  1. Delete the placeholder throw in the Java starter. 0.05 h · ROI 5–20 marks/hour.
  2. Pre-solve the 9 observed frontend tasks. 3 h · beats every topic under every assumption.

The library

Core

File What
OBSERVED-POOL.md ⭐ All 27 real tasks, cleaned, solved, verified. The most valuable file here.
03-ROUND-STRATEGY.md Attempt order, 60-minute clock, per-part playbook, IDE facts
02-TOPIC-WEIGHTAGE-ANALYSIS.md What is worth studying, with the numbers
91-CHEATSHEET.md One page for the last hour
99-STUDY-PLAN.md 7 / 14 / 30-day plans
90-RESOURCE-LIBRARY.md External practice, link-checked

Backend / DSA — Java

130 practice problems, every one solved and executed.

File Share of the backend mark
problems/00-BACKEND-OVERVIEW.md —
solutions/java/JAVA-TRAPS.md the template trap + 10 verified gotchas
problems/P01-strings.md 24.4% — string → structure
problems/P02-counting.md 21.1% — hash-map frequency
problems/P03-accumulate.md 19.2% — single-pass accumulate
problems/P04-order-statistics.md 13.7% — count smaller/larger
problems/P05-sliding-window.md the dated one — a sliding-window question, 13 Oct 2025. 20 problems
problems/P06-sorting-searching.md named in the experience set. 16 problems
problems/P07-dynamic-programming.md Decode Ways, 9 Oct 2025. 13 problems — time-boxed, worst ratio in the library
problems/P08-stacks-queues.md "queues" named, never captured. 15 problems + 6 executed gotchas

SQL

File What
sql/SQL-00-FROM-ZERO.md The whole clause set, taught from nothing
sql/schemas.sql · sql/queries.sql 31 tables; the 10 observed tasks solved
sql/practice.sql 16 drills

Frontend

File What
frontend/FE-00-THE-FORMAT.md Edit-don't-write, the five rules, the API you need
frontend/F1…F9/ 9 runnable starter + solution pairs
frontend/learn/00-INDEX.md ⭐ HTML/CSS/JS from zero — 8 chapters, 12 h. Start here if you write Java and nothing else.
frontend/learn/L08-exercises.md 12 original exercises in the exam's format, with solutions and the 8 traps
frontend/learn/exercises/E01…E12/ 12 runnable starter + solution pairs

Mock

File What
mock/MOCK-60MIN.md Timed paper — original problems in the observed format
mock/MOCK-60MIN-SOLUTIONS.md Solutions, traps, and where to study each miss

Evidence and method

File What
research/00-source-inventory.md Every source, what it gave, what was unreadable
research/01-local-corpus-findings.md The 27 tasks analysed
research/02-online-research-log.md 21 network operations, credibility-rated
research/03-pattern-synthesis.md Conflict table — where sources disagree
research/04-qa-report.md What QA found and fixed, including my own errors
data/analyze.py · data/weightage.json · data/analysis-tables.md The model. Re-run the script and every number regenerates.
_backlog/OTHER-ASSESSMENTS.md Everything found about Assessments #1 and #4 — parked, not developed
PROGRESS.md Build log

What the evidence says this exam is

Corpus 27 distinct real tasks, 46 independent captures
Frontend 9 tasks. The pool repeats verbatim — 6 questions captured in two independent sessions three weeks apart. Plain HTML/CSS/JS; zero React, zero Vue.
SQL 10 tasks. MySQL. Never more than 3 tables joined out of 7–8 shown. 7 of 10 turn on a boundary or a NULL. No window functions, no CTEs, no triggers.
Backend 8 tasks. Function-only — no main, no Scanner. Arrays, strings, hash maps. Nothing needs better than O(n²).
Scoring Per test case, with hidden tests. There is a score cutoff, not a "finish 2 of 3" cutoff.

Everything here has been executed

cd solutions/java && javac BackendObserved.java && java BackendObserved   # 25 assertions
cd problems/java  && javac P01Strings.java && java P01Strings             # 45
                     javac P02Counting.java && java P02Counting           # 38
                     javac P03Accumulate.java && java P03Accumulate       # 16
                     javac P04OrderStats.java && java P04OrderStats       # 15
                     javac P05SlidingWindow.java && java P05SlidingWindow   # 43
                     javac P06SortSearch.java && java P06SortSearch         # 28
                     javac P07DynamicProgramming.java && java P07DynamicProgramming  # 42
                     javac P08StacksQueues.java && java P08StacksQueues     # 41
cd mock && javac M1Warehouse.java && java M1Warehouse                     # 6
python3 sql/run_sql.py                 # 10 observed queries
python3 sql/run_practice.py            # 16 practice drills
python3 frontend/verify_frontend.py         # 50 checks in real Chrome
python3 frontend/learn/verify_lessons.py    # 46 lesson claims, in real Chrome
python3 frontend/learn/verify_exercises.py  # 82 exercise checks + doc/code agreement
python3 mock/verify_mock.py                 # 9 checks
python3 data/analyze.py                     # regenerates every number

299 Java assertions · 26 SQL queries · 187 real-Chrome browser checks. No solution ships unexecuted. Last full re-run: 2026-09-17, everything green.

Honest limits

  • Backend n=8 is the thinnest evidence base. The split within the backend mark is directional, not precise.
  • The frontend pool size is unknown. We see 6 questions twice; the true pool could be larger. The model gives a scenario range rather than inventing a number.
  • B1's rule is reconstructed — its statement was never captured. Re-read the real one.
  • The throw trap is single-source, though the failure mechanism is verified by compiler output and the insurance costs three minutes.
  • 9 of 27 tasks come from coaching notes, one credibility tier below exam-screen captures.
  • C++ availability is unresolved — and irrelevant to you, since you write Java.