01
Questions to say aloud
- 一次面接では学食のプロジェクトを伺いましたが、指標を変えることに反対された方はいましたか。どのように合意されたのか教えてください。
- なぜ、同業他社ではなく当社のデータ基盤職なのですか。
- 香港大学で学んだ情報システムの内容を、青凪デジタルのデータ基盤職で最初にどう生かせると考えていますか。
Interview Preparation
Generated from your profile, application record, submitted content, and company materials. Verify company facts and selection dates against official sources.
Run through this on the way — flow, self-intro, strengths to land, questions to ask back, and the day-of checklist.
StrategyThis interview needs to prove one thing: the method for turning a metric definition into a team agreement is repeatable — and the pasted job posting assigns exactly that work to the data-platform role.
香港大学で情報システムを学んでおります、留学太郎と申します。学部では学食の混雑を可視化する学生プロジェクトに18週間取り組み、約2万4千件の匿名記録を整理いたしました。当初は曜日別の平均待ち時間を見ておりましたが、平均ではピークが隠れると分かり、15分単位の待ち時間の中央値へ指標を統一することをチームに提案し、合意のうえで表示の試行まで進めました。この経験から、分析そのものよりも、関係者が同じ定義で判断できる状態をつくる仕事に関心を持っております。募集要項を拝見する限り、貴社のデータ基盤職は指標の定義を整え、利用部門と合意するところまで担当すると理解しております。卒業は2027年6月、日本語はJLPT N1です。本日はどうぞよろしくお願いいたします。The four-step method for turning a definition conflict into an agreement is repeatable.
Only what was verified is called a result; everything else is named as unverified.
The habit of writing definitions down so the next person can take over.
データ品質の基準を利用部門と合意される際、定義の変更はどのような手順でレビューされていますか。
The posting lists agreeing data-quality standards with user teams as a responsibility. This question reads that line as a daily action and connects straight to your own evidence.
入社後三か月のあいだに、既存の指標定義をどの順番で読み込むことを期待されていますか。
It asks how the first stretch after joining actually goes, which shows you read the job description far better than asking about company culture.
新卒の方が指標の定義変更を提案された実例があれば、どこが一番難しかったかを伺えますか。
It lands on a real difficulty and, at the same time, tests whether new graduates here are allowed near definitions.
Confirm the flow, common humanities/STEM follow-ups, and how company research becomes motivation and reverse questions.
They check whether you can explain yourself in a minute, and whether pace, eye contact, and structure hold.
They check whether the judgment inside the story is yours, and whether the motivation survives swapping the company name.
The weight of this round: how you reached agreement with someone who disagreed, and how clearly you explain a trade-off.
They check whether you read the job description and whether the people present can answer what you ask.
学食プロジェクトの18週間で、最も自分で判断した一手は何ですか。 / 香港大学での情報システムの学びとJLPT N1を、青凪デジタルのデータ基盤職にどう接続しますか。 / 入社後三か月で、募集要項の「データ品質の合意」に関わる前に何を読み、誰に確認しますか。
15分単位の中央値へ変える際、過去比較を心配した相手と、どの二案をどう並行させましたか。 / 三つの待ち時間の定義を一枚にして次年度へ渡したとき、誰が次に使える状態をどのように確認しましたか。
曜日別平均がピークを隠すと判断した根拠を、約2万4千件の匿名記録の扱いとともに説明してください。 / 平均値と15分単位の中央値を二週間並行させた比較で、何を確認し、何は確認していませんか。 / 貼付の募集要項にあるデータ接入から利用部門への提供までの責任を、食堂プロジェクトのどの一往復で示せますか。
Take one responsibility sentence from the pasted posting and connect it to the part of the cafeteria project where you unified the metric.
Rewrite the responsibility sentence as a question about action: who performs that duty day to day, and when.
Work backwards from the posting's description of the first months to decide which definition documents to read and whom to meet.
Look at how much of the role is agreement with user teams and judge whether that is the intensity of collaboration you want.
Quote the source sentence once and say where you read it; hedge everything else with 「拝見した範囲では」.
Reread the ES you submitted to this company and mark the three sentences most likely to be probed.
Check how many days transcripts and enrollment certificates take to issue, and request them early if needed.
Copy each responsibility from the pasted posting into a list and keep only sentences you can quote verbatim.
Record yourself saying the self-introduction and three frequent questions aloud, and check only where the conclusion sits.
Hang the suit the night before, check shirt and shoes, and keep hair off your eyes.
Confirm the route and transfers, and set departure for 30 minutes before arrival.
Arrive 10 minutes early and give your university, full name, interview time, and the contact's name.
Do not chat with other candidates, silence the phone and put it in your bag, and sit without leaning back.
Knock two or three times, say 「失礼いたします」 after the reply, close the door quietly, then bow.
Sit only when invited, place your bag beside the chair on the outer side, and rest both hands on your knees.
When you cannot catch something, say 「恐れ入りますが、もう一度お願いできますでしょうか」 rather than answering on a guess.
Stand and thank them, turn at the door and bow once more, say 「失礼いたします」, and close the door quietly.
Answer anchor:香港大学で学んだ情報システムの内容を、青凪デジタルのデータ基盤職で最初にどう生かせると考えていますか。 They are checking whether working in Japan is a concrete choice or an extension of liking the country.
Answer anchor:募集要項にある「利用部門とデータ品質を合意する」という責任を、学食プロジェクトのどの経験につなげますか。 They are checking whether the motivation survives swapping the company name.
Answer anchor:18週間・約2万4千件の匿名記録を扱った学食プロジェクトで、平均値ではなく15分単位の中央値を選んだ判断を説明してください。 They are checking independence of judgment and whether the method repeats.
Answer anchor:売上データを扱わず、アンケートと現場観察だけで試行を確かめた範囲を、入社後の仕事にどう持ち込みますか。 They are checking whether you have actually looked at the product and can offer a grounded observation.
Answer anchor:2027年6月卒業見込みとJLPT N1を、入社後三か月で既存の定義資料を読む計画にどう結びつけますか。 They are checking whether you have thought about residence status and career direction rather than avoiding it.
Answer anchor:自社プロダクトのデータ基盤職を選ぶという就活の軸を、青凪デジタルを含む応募先の判断にどう使っていますか。 They are checking consistency of your job-search axis, not the number of applications.
Align the basics: industry, company type, role, daily work, valued skills, and where your experience connects.
IT and web firms earn in three ways: building systems for others (contract work), selling their own product, and renting out engineers by the month (SES). Recent growth sits with product companies and data preparation, because the data is already stored while very little of it is usable by the business as it stands.
The pasted posting says this is development and operation of the company's own product, that the role sits in the data-platform group, and that the data comes from that product — which reads as a product company, not contract development and not SES. None of the SES red flags appear in the posting: no client-site placement, no undecided assignment. If placement or dispatch comes up during the interview, confirm the contract form on the spot.
A data-platform engineer, in one line, is the person who makes data safe for other people to use. The common misreading is 'builds reports' or 'builds machine-learning models'; the actual work is intake, definitions, quality monitoring, and delivery, while models belong to someone downstream.
Most of a day goes to three things: fixing the pipeline that failed overnight, answering a user team asking why a number differs from last week, and connecting a new source while writing down its definition. Across a year the shape is: three months reading existing definitions and code, then ownership of one data domain, then joining reviews of definition changes.
Three parts of the cafeteria project map onto the first three: seeing that a weekday average hid the peak (definitions), convincing the team to switch to a 15-minute median (communication), and organizing about 24,000 records into a working pilot (fundamentals). What is missing is production operations experience — concede that directly in the answer instead of covering it with the course project.
Subtract the industry's common ground → find this firm's distinction → your experience matters precisely there.
The whole industry talks about data utilization, DX, and building in-house. Those three phrases fit any employer, so on their own they are not a reason.
Two distinctions are readable in the pasted posting. First, the data-platform role owns intake through delivery in the first person rather than being split by layer. Second, the responsibilities explicitly include agreeing data quality with user teams, which puts negotiation inside an engineering job. Company facts outside the posting are not asserted.
Those two lines are exactly the part where you have completed one full loop: cleaning records to a usable state alone, then persuading the users to change the definition. In a team split by layer that experience counts as half a story; in a role that states first-person ownership, it is one continuous story.
Four beats: concede that rivals also do data preparation and do not pretend it is unique → point out that this firm wrote agreement with user teams into an engineering role, with a quotable line → say that treating negotiation as part of the work is the definition you want → show with the cafeteria project that you have already walked that line once. Leave the exact wording to the room; do not memorize it as a script.
'Because I like Japan' and 'I want to grow in Japan' fit every Japanese employer, so they add nothing. Turn language and an outside perspective into concrete actions: N1 lets you read definition documents directly and take part in definition reviews, and reading English source material is how you found the basis for choosing a median. Present them as skills used at work, not as an identity label.
◎ = frequent/required this round, ○ = possible, △ = rare. Keep answers consistent with what you already submitted.
The core of a second interview: how you moved a disagreement to an agreement when someone pushed back.
Submitted-content anchorThe submitted ES says the metric was unified to the median wait time in 15-minute windows. Keep the spoken answer on that exact claim and do not upgrade it to a shorter wait time.
Follow-ups並行して見た2週間のあいだに、判断が変わった点はありましたか。 / 反対されていた方は、最終的に何に納得されたのでしょうか。
AvoidRisk:Saying the pilot actually shortened the wait; no sales data supports that sentence. / Skipping the agreement with 'everyone came round in the end' — that step is exactly what this round examines.
Checks whether the motivation still holds after the company name is swapped out.
Submitted-content anchorThe ES motivation paragraph says you want work that lets other people use data with confidence. Keep the spoken version pointing the same way instead of introducing a new phrasing.
Follow-upsその工程を担当するうえで、いまご自身に一番足りないと思われる点は何ですか。 / 同じ職種で他社も受けていらっしゃると思いますが、決め手はどこにありますか。
AvoidRisk:Using an unsourced judgment such as 'your company leads the industry' as the reason. / Stating company facts that are not in the pasted posting as if they were known.
Checks that you know the edge of your own experience and that your ramp-up plan is concrete.
Follow-ups読む順番をご自身で決めるとしたら、何から手をつけますか。 / 分からないことが出てきたとき、誰にどう聞きますか。
AvoidRisk:Using the course project as a stand-in for production experience. / Saying only that you will study hard, with no order and no named counterpart.
Checks your working style and whether your self-description matches the evidence.
Follow-upsまとめ役に回れない場面では、どう動きますか。 / その進め方でうまくいかなかった経験はありますか。
AvoidRisk:Calling yourself a leader without one example of someone changing what they did. / Describing the role as universally useful and never naming its cost.
Checks whether you actually read the job description and whether the people in the room can answer your question.
Follow-upsその点について、ご自身ではどうお考えですか。
AvoidRisk:Asking for public information you could look up, such as headcount or founding year. / Opening with pay and overtime before asking anything about the work itself.
Motivation / reverse questions / post-join vision / 'will you join' all point to the same anchors.
The closing line of the self-introduction, where the motivation lands, both reverse questions, and the answer to 「内定が出たら入社されますか」 must point at the same anchors. To check: write each of those four as one sentence and see whether 'definition' and 'user' both appear. If one of them is left with only the company or personal growth, that is the drift.
We can't assert these company facts for you. Verify before the interview or ask on the day.
The interviewers' teams and how many will attend are unknown: ask HR beforehand, or confirm their teams at the opening.
The length of the interview and whether technical questions are included are unconfirmed, and both change how you budget the self-introduction and reverse questions.
Whether the assignment is fixed and whether the role is split into courses must be confirmed with the company before discussing the first year.
Company facts outside the pasted posting — client mix, team size, technology stack — have no source, so do not assert them in an answer.
If residence status and start date come up, describe only the current situation and the window you will confirm; the procedure itself follows the Immigration Services Agency's current notices.
EVIDENCE · JUDGMENT · ACTION
Scoring methodFour 0–5 checks assess whether the preparation evidence is usable. This is not an interviewer score, an aptitude score, or a probability of an offer. 0 means the material is missing; 3 means the source exists but verification or rehearsal remains; 5 means the source is traceable, the answer has been rehearsed aloud, and the current interview detail is confirmed.
Total12/20
| Dimension | Score | Evidence for the score | What it changes |
|---|---|---|---|
| Traceability of experience evidence | 4/5 | The ES supplies the duration, record count, metric choice, and validation boundary; no independent record shows the effect after the pilot. | Say that the pilot and handover were completed, not that the business outcome improved. |
| Connection to role responsibilities | 3/5 | One responsibility can be quoted from the pasted text, but it has not yet been compressed into the last two sentences of the motivation answer. | Quote the responsibility once, then say which part of the project maps to it. |
| Spoken response under follow-up | 3/5 | The four STAR steps are drafted; there is no 90-second recording or review of one challenge about opposition. | Practice why the metric changed and who objected, then how agreement was reached. |
| Conditions for this interview | 2/5 | The duration, interviewer mix, and technical-question scope are not confirmed. | Confirm these with HR before deciding the length of the answer or the reverse questions. |
ActionMake a one-page evidence card for the median decision and record a 90-second Japanese answer. Include the disagreement, the applicant's decision, parallel validation, handover, and the unverified boundary.
DeliverableOne evidence card and one recording no longer than 90 seconds.
Done whenThe recording does not describe surveys or onsite observation as a business outcome, and it can answer one follow-up about who objected and how the applicant responded.
ActionCopy one responsibility sentence from the latest role text and rewrite the last two sentences of the motivation answer: state the responsibility first, then connect it to the applicant's project decision.
DeliverableA responsibility excerpt with its location or version date, plus a two-sentence Japanese close.
Done whenThe two sentences use only the pasted text and the applicant's experience, with no unverifiable claim such as 'an industry leader.'
ActionConfirm duration, interviewer department and number, and technical-question scope with HR. Write three columns: confirmed, unconfirmed, and to ask in the room.
DeliverableOne confirmation sheet for this interview.
Done whenUnknown items are still marked unknown, and the applicant has used the sheet to decide the self-introduction length and one reserve reverse question.
Aonagi Digital Co., Ltd. is a fictional employer used for this sample, and every company fact here comes from the fictional posting; none of it corresponds to a real company.
The interview flow and the time given to each stage are general guides; the actual schedule is set by the employer and needs confirming on the day.
The industry and role descriptions come from general tendencies in the capability profile and vary by company and by year.
Interview prep is not a standalone report. Carry it back into ES, experience materials, company decoding, or your application tracker.