Practice

Workplace simulations and interview practice. Optional beyond the weekday lessons; studios embed selected scenarios. Everything simulated is labelled.

100scenarios
25capabilities covered
62 hof simulated work
150interview questions

Scenarios by capability100

S01 · 5 scenarios
Production incident
- The load forecast that moved its peak by three hours35 min
- A renamed API field quietly sets competitor fares to zero40 min
- A competitor price feed that passes every health check35 min
- Reconfirmation calls that went out twice35 min
- The assistant that stopped finding answers overnight40 min
S02 · 4 scenarios
Senior code review
- Review a churn model with a 0.97 AUC40 min
- Review a network revenue query that NULLs quietly bend30 min
- Review a nightly no-show scoring script before it ships35 min
- Review a "busiest flight per route" query that looks right35 min
S03 · 4 scenarios
Stakeholder simulation
- Three stakeholders and one overbooking limit40 min
- Mortgage assistant, speed versus safeguards35 min
- "95% accurate per flight": negotiating a forecast target35 min
- Catering wants the exact meal count three days out35 min
S04 · 3 scenarios
Messy real-world data
- Claims fell 75% - or did the labels just not arrive yet?35 min
- Booking events that arrive twice, late and out of order35 min
- A raw flight table and a Friday deadline40 min
S05 · 3 scenarios
ADR centre
- ADR: batch or online scoring for no-show predictions35 min
- ADR: one global demand model or one model per route35 min
- ADR: how a forecast should be refreshed after the market changes40 min
S06 · 5 scenarios
Live coding and SQL
- SQL gym: rolling and same-weekday features with window functions30 min
- Python gym: entity-safe lag and rolling features in pandas30 min
- SQL gym: consecutive nearly-full streaks and why the first query lies35 min
- Python gym: turning cumulative booking curves into pickup in pandas35 min
- SQL gym: booking pace against the same weekday last year40 min
S07 · 4 scenarios
ML system design
- System design: a demand forecasting platform for 2,000 route-days45 min
- System design: card-fraud scoring inside a 50 ms budget45 min
- System design: ranking ancillary offers during online check-in45 min
- System design: catching wrong fares before they sell40 min
S08 · 6 scenarios
Ambiguous business case
- Conversion dropped. Fix it.30 min
- IST-ADB is down 5%: cut fares 10%?35 min
- An LLM support assistant with no success criteria40 min
- Add a third daily frequency on IST-LHR?35 min
- How many contact-centre agents does the holiday peak need?35 min
- NPS fell six points. What happened?30 min
S09 · 3 scenarios
Verbal defense
- Defend a booking forecast under challenge30 min
- Defend a loan-default model to the risk committee35 min
- Defend a delivery-fee experiment before it is trusted35 min
S10 · 3 scenarios
Statistical traps
- A loyalty campaign readout with three statistical traps35 min
- The checkout test that was significant on day 635 min
- The rescue package that rescued 87% of weak flights30 min
S11 · 3 scenarios
Ablations
- Which part of forecast v2 actually helped?40 min
- Which parts of the v3 booking forecast earn their place?40 min
- Is the new assistant pipeline better, or luckier on 60 questions?35 min
S12 · 4 scenarios
Decision theory and value
- 1,000 retention calls - who gets one?35 min
- How many meals to load, when running short costs more than waste?35 min
- What is an 8% lower forecast error worth to a parts warehouse?35 min
- Auto-resolve or review: a confidence threshold with 1,000 reviews a day40 min
S13 · 3 scenarios
Robustness and stress tests
- Stress-test a parcel delay model before peak season35 min
- Stress-test the booking forecast before its feed migration35 min
- Break the fare recommender before it prices a flight35 min
S14 · 3 scenarios
Paper reproduction
- Check a forecasting foundation-model claim40 min
- Check an LLM-judge agreement claim before adopting it35 min
- Does the new optimizer really beat AdamW?35 min
S15 · 4 scenarios
RAG evaluation workbench
- Choose a RAG configuration for a fare-rules assistant40 min
- Where should the abstention threshold of a baggage-policy assistant sit?35 min
- A retrieval upgrade scores worse because the gold labels are incomplete35 min
- Turkish questions against English policy pages40 min
S16 · 5 scenarios
LLM failure lab
- Unsupported citations and broken JSON in a claims assistant35 min
- Injected instructions and a looping helpdesk agent40 min
- A claims summariser that never saw its last document35 min
- A refund assistant that paid twice, in the wrong unit35 min
- An oversized voucher late in a long baggage-claim conversation40 min
S17 · 4 scenarios
Multi-agent company
- Run an overbooking model through the whole company45 min
- Add a third London frequency? Run it through the company45 min
- A vendor no-show model trial that wants the whole booking record40 min
- Delay notifications through the whole company40 min
S18 · 5 scenarios
End-to-end lifecycle
- A churn model from intake to retirement45 min
- A route forecast from intake to handover50 min
- A procedures assistant from intake to retirement50 min
- A cancellation-risk model whose success rewrites its own labels50 min
- A case-summary assistant from intake to model retirement45 min
S20 · 4 scenarios
Golden test set and regression
- Golden set regression for a refund-email extractor40 min
- A holiday-aware rewrite of a demand baseline and its golden cases40 min
- The golden answers went stale before the retriever did40 min
- A directional golden case fails on a good demand forecaster40 min
S22 · 4 scenarios
Hiring committee
- Final hiring committee on your own evidence45 min
- Committee review of a simulated analyst candidate pack40 min
- A forecasting headline under committee questioning40 min
- Four interviewers, one must-have, no consensus35 min
S23 · 4 scenarios
AI cost–quality
- Cut LLM cost without losing quality40 min
- How many chunks to retrieve, priced per correct answer40 min
- Self-hosting a quantized model against an API, with gates40 min
- Batch pricing is cheap until the deadline arrives35 min
S24 · 3 scenarios
Agents vs workflows
- Fuel invoice reconciliation, agent or workflow40 min
- Load-factor anomaly tickets, runbook or investigating agent40 min
- Month-end commentary, a ten-step loop against one call35 min
S27 · 4 scenarios
Teach-back and Socratic interviewer
- Teach back calibration to a new analyst25 min
- Teach back why fluent LLM output can be wrong25 min
- Teach back why a positive fare coefficient is not a pricing result25 min
- Teach back what an 80% interval promises and how to check it25 min
S31 · 5 scenarios
Executive presentation
- Executive memo on a forecast pilot35 min
- Readout memo for a fare-rules assistant pilot35 min
- Briefing the network team on a competitor capacity step40 min
- The revenue dashboard that reads 3.7% low35 min
- Executive brief after an agent paid refunds twice35 min
S32 · 5 scenarios
Research reading protocol
- Read a churn paper before believing its abstract35 min
- Read an agent benchmark paper before adopting its framework35 min
- Does long context make retrieval obsolete? Read the claim40 min
- A language-model judge that agrees with humans 85% of the time35 min
- LoRA matches full fine-tuning: how much can 400 invoices show?35 min

Capabilities that are workflows, not scenarios

Interview gym159

Today's random question

13 questions

Business judgment
Before anyone builds it, the head of marketing asks you: "Is a churn model worth it for us?" The company has 1 million subscribers and about 2% monthly churn.
senior15 min
Business judgment
A manufacturing plant uses a model that outputs the probability a part is defective. Inspecting a part costs 20; shipping a defective part costs 500 (warranty and rework).
mid8 min
Business judgment
The chief commercial officer asks: "Should we launch a cheaper basic fare without a checked bag on our domestic routes?" You have 20 minutes. How do you structure your answer?
senior20 min
Business judgment
You lead a team of six. Leadership hands you eight AI/ML project requests for next year (a GenAI customer assistant, demand forecasting upgrade, fraud model refresh, dashboard automation, and four…
staff25 min
Business judgment
Over a year, your team's recommendation model improved click-through rate by 15% in successive A/B tests, but revenue per visitor is flat and returns increased.
staff20 min
Business judgment
A marketing email campaign cost 81,000 and was sent to 450,000 customers, with 50,000 comparable customers held out at random (hypothetical figures).
senior15 min
Business judgment
Nine months into a demand-forecasting rebuild, the team has reached 70% of the accuracy-improvement target. The sponsor says "we have already spent 1.2 million; we must finish".
staff25 min
Business judgment
A team proposes a forecasting project. Hypothetical estimates: build cost 250k, run cost 90k per year, benefit of 400k per year if it works, with a ramp to half the benefit in year 1, and a 60%…
mid10 min
Business judgment
Leadership sets the goal "reduce route-level forecast MAPE by 20% this year" and ties bonuses to it. After six months the average MAPE across 200 routes (unweighted) dropped from 21.3% to 16.6%.
senior15 min
Business judgment
Your team of five engineers (about 100 productive engineer-days per month) operates 14 production models, each costing around 4 engineer-days a month in monitoring, retraining and incident work, so…
staff25 min
Business judgment
A four-week pilot of a pricing-recommendation model on 20% of routes shows revenue +1.1% against control, with a 95% interval of -0.3% to +2.5%.
senior15 min
Business judgment
The CFO asks you to report the annual business value created by the company's ML systems. Five teams each submit a number from their own A/B tests: 6.0, 5.2, 4.1, 3.0 and 2.2 million (hypothetical).
staff25 minin review
Business judgment
A retention programme calls customers ranked by an uplift model. A holdout experiment on 10,000 customers (hypothetical, 1,000 per decile of predicted uplift) shows the incremental number of…
senior15 min