Upstox Data Analyst Interview Preparation

Define whether a row represents an order, item, customer or event. A one-to-many join can multiply a measure stored at a coarser grain. Aggregate deliberately and reconcile a sample by hand. Record how returns, cancellations and missing joins affect the metric.

Check your understanding: Verify a fixture with one order and several line items.

Common trap: Summing a repeated order total after a one-to-many join.

Concept reference: PostgreSQL: reading EXPLAIN plans

For a skewed distribution, the median describes the middle observation without being pulled as strongly by a few extremes. The mean remains useful for totals and expected value questions. Explain the business question and show the distribution instead of declaring one statistic universally better.

Check your understanding: Compare mean, median and total before and after adding one large transaction.

Common trap: Discarding outliers without investigating whether they are real.

Concept reference: Python: statistical functions

Measure the expensive query with realistic parameters and inspect its execution plan. Check estimated versus actual rows, join cardinality and filtering before adding indexes. Keep business-result checks alongside performance checks so a faster query does not silently change the metric.

Check your understanding: Compare both results and execution work before and after the proposed change.

Common trap: Optimizing latency while accidentally changing the denominator.

Concept reference: PostgreSQL: reading EXPLAIN plans

First, identify technical constraints and define measurable service objectives. Next, connect the objective to data preparation, evaluation and production feedback. Contrast architectural trade-offs across simplicity, correctness, maintainability and scale, explicitly mitigate the risk of shared-IP limits reject legitimate users while credential attacks continue, and confirm system stability using separate identity and network limits plus an abuse-simulation report.

Common trap: Reaching for a specific library or framework before defining constraints, failure envelopes, and automated verification criteria.

Prioritise the failure mode exhibiting the highest user blast radius and lowest observability. Formulate an explicit containment boundary, implement idempotent retries with jitter, and establish an automated rollback threshold. Verify resilience through a server-authoritative timestamp trace and boundary property tests.

Common trap: Relying on passive monitoring dashboards without defining explicit error-budget alerts, rollback triggers, and verified recovery procedures.

Structure the response using Context-Decision-Tradeoff-Result: articulate the business and technical constraints, compare viable alternatives, explain the implementation (connect the objective to data preparation, evaluation and production feedback), and document the accepted trade-off. Provide concrete proof: a project example, measured result and repeatable verification step.

Common trap: Speaking only in high-level abstractions or team accomplishments without detailing your direct implementation decisions, trade-offs, and measured results.

Formulate a falsifiable hypothesis from observable telemetry before altering configurations. Then compare data slices, leakage checks, baseline metrics and production feedback distributions. Isolate the defect to the smallest reproducible boundary, validate root cause with evidence, and confirm full resolution using segmented service-level indicators, an exemplar trace and an alert threshold.

Common trap: Applying speculative fixes or restarting services blindly without establishing an observable signal connected to a falsifiable hypothesis.

Produce a reproducible evaluation notebook with slice metrics and a monitoring threshold. Document baseline assumptions, technical mechanism (connect the objective to data preparation, evaluation and production feedback), rejected alternatives, bounded failure envelopes, and deterministic pass criteria. Supply reproducible verification via shadow-traffic comparison and an error-budget based cutover rule.

Common trap: Presenting architecture diagrams or slides lacking automated unit/integration tests, observable metrics, or automated rollback configurations.