Question 1

Azure Fabric in Cloud Architecture?

Accepted Answer

**Section 1 — The Context (The 'Why')**
The primary challenge for this design in Cloud/Tools is balancing scale, cost, and reliability. At scale, naive approaches fail: single points of failure cause cascades, schema evolution breaks consumers, and over-provisioning explodes cost. Failure modes include silent data loss from non-idempotent writes, cascading job failures from tight coupling, and operational burden from manual intervention....

Question 2

Azure Functions vs. Logic Apps?

Accepted Answer

Why choose: Different abstraction levels—code vs. workflow. Functions: code-first, serverless; custom logic, APIs, event-driven. Logic Apps: low-code, 400+ connectors; designer-driven. When Functions: complex transforms, code reuse, fine-grained control. When Logic Apps: orchestration with minimal code, SaaS integration, non-devs modifying flows. Scalability: Functions scale with demand; Logic Apps have execution limits (e.g., 90-day run history)....

Question 3

Agile in project management?

Accepted Answer

**Agile**: Iterative development in sprints. Collaboration and adaptability. **Roles**: PO, Scrum Master, team. **Ceremonies**: Planning, standup, review, retro. **Data deliverables**: Pipelines, datasets, quality....

Question 4

Find orders exceeding $1,000 in the last 30 days.

Accepted Answer

**SQL**: WHERE amount > 1000 AND order_date >= CURRENT_DATE - INTERVAL '30 days'. Index on (amount, order_date). Parameterize dates for testing. For streaming: watermark + window agg.

Question 5

Find top 3 products sold based on total quantity.

Accepted Answer

**SQL**: GROUP BY product, SUM(quantity), ORDER BY total DESC LIMIT 3. Or RANK() OVER (ORDER BY SUM(quantity) DESC) with filter rk <= 3. RANK handles ties (same rank); DENSE_RANK for consecutive ranks.

Question 6

List customers with more than 5 orders.

Accepted Answer

**Why HAVING vs WHERE:** WHERE filters before aggregation; HAVING after. For 'more than 5 orders,' you must aggregate first—HAVING is correct. Using WHERE with a subquery is an alternative but less clear.

**Scalability:** (1) Index customer_id, order_id—enables index-only scan for COUNT. (2) For billions of rows: pre-aggregate in a summary table (customer_order_counts) refreshed incrementally....

Question 7

Python list operations.

Accepted Answer

**Core Operations:** Indexing, slicing (lst[i:j:k]), append, extend, insert, remove, pop, sort, reverse. Membership (in) is O(n); for repeated lookup use set.

**Gotchas:** lst.reverse() and lst.sort() mutate in-place, return None. reversed(lst) and sorted(lst) return new iterable/list. lst += [1] mutates; lst = lst + [1] creates new. Slice assignment lst[1:3] = [a,b] can change length.

**Production:** Prefer list comprehensions over manual loops. Use bisect for sorted lists....

Question 8

Unix scripting in data engineering?

Accepted Answer

**Why Unix in Data Eng:** File staging, cleanup, lightweight transforms before Spark. find, awk, sed, xargs, cron—ubiquitous, no runtime deps.

**Use Cases:** find /data -name '*.csv' -mtime +7 | xargs gzip. awk -F',' '{print $1,$3}' for column extraction. sed for bulk replace. Cron for scheduling: 0 2 * * * /scripts/ingest.sh.

**Scalability:** Shell for single-node, pre-processing. For distributed: Spark. Best practice: set -e (exit on error), validate paths, log to file....

Question 9

Convert row-level records to column records.

Accepted Answer

**Architectural Logic**: Pivoting (rows→columns) is a reshape operation with cardinality and performance implications. **SQL**: Aggregate + CASE: `SELECT product_id, MAX(CASE WHEN month='Jan' THEN sales END) AS jan_sales, ... FROM sales GROUP BY product_id`. Or PIVOT (SQL Server). **Spark**: `df.groupBy("product_id").pivot("month").agg(sum("sales"))`. **Why Cardinality Matters**: High cardinality in pivot column (e.g., pivot by user_id) causes column explosion—hundreds of columns, schema bloat....

Question 10

ER Modeling vs. Dimensional Modeling?

Accepted Answer

**Architectural Logic**: ER and dimensional serve different purposes—transactional integrity vs analytical performance. **ER**: Normalized; entities and relationships; OLTP; minimizes redundancy; supports transactional integrity. Many tables; complex joins. **Dimensional**: Star/snowflake; facts and dimensions; OLAP; optimized for analytics. Denormalized; fewer joins; intuitive for business. **Why Both**: ER for operational systems; dimensional for warehouses....

Comcast Data Engineer Interview Questions

Difficulty Breakdown

Key Topics Covered

How to Use This Guide

Companies asking these questions

All 18 Questions

More Interview Prep Guides

Unlock All Expert Answers