How to Crack Amazon's SQL Interview: Key Skills and Strategies

·1 views

Amazon's SQL interviews for roles like BI Engineer, Data Analyst, and Data Scientist test more than syntax — candidates are evaluated on logical thinking, communication, and handling messy real-world data. Interviews typically take the form of a technical screen or a whiteboarding case study, with interviewers looking for correctness, clear reasoning, and structured query writing. Core topics include JOIN types, window functions such as ROW_NUMBER and LAG, aggregations with CTEs, and date-based calculations like rolling averages and cohort analysis. Candidates are advised to restate the problem, clarify ambiguous metrics, and build queries in logical CTE blocks rather than complex nested subqueries. Interviewers often embed intentional data traps — such as nulls in join keys or duplicate rows — to test whether candidates spot and address data quality issues proactively.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Why filtered data should be labeled, not deleted, in software systems

A software developer building a mileage-tracking app discovered a critical flaw months after shipping a GPS data filter: rejected readings were being permanently deleted, making it impossible to verify whether the filter was working correctly. The fix involved replacing hard deletions with labeled categories, so that implausible GPS points were marked and stored rather than discarded, with only physically impossible values still removed entirely. This approach allowed disputed expense claims to be resolved by showing drivers exactly which readings were excluded and why. Tunable thresholds became possible because engineers could replay real journeys and adjust filtering logic based on retained data. The developer argues this principle applies broadly to any system that rejects data — from fraud scoring to log pipelines — since deleting a record destroys the only evidence needed to audit the decision.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Ventoy Lets You Boot Multiple ISOs From One USB Drive Without Reformatting

Rufus and Ventoy are both tools for creating bootable USB drives, but they work in fundamentally different ways. Rufus writes a single ISO directly to a USB drive, typically requiring the process to be repeated whenever a different operating system is needed. Ventoy, by contrast, is installed once on the USB drive, after which users can simply copy and delete ISO files as needed without reformatting. This allows a single large USB drive to store multiple operating systems and recovery tools simultaneously, accessible via a boot menu. Ventoy also supports a persistence feature on select Linux distributions, enabling users to retain system changes across reboots when running a live environment.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Developer Builds Open-Source Mobile Terminal App Using the App Itself

A developer has created Termish, an open-source mobile tool designed for remote work via SSH/Mosh terminals, file management, and AI-assisted development. Unusually, much of Termish was built using Termish itself, with the developer working from a phone connected to a host machine running herdr and OpenAI's Codex. The app targets a workflow gap where developers want to start, inspect, or continue coding tasks from a phone without replacing a desktop setup entirely. Termish integrates terminal access, AI agent interaction, screenshot uploads, Git diff reviews, and remote screen viewing into a single touch-optimized interface. The source code is publicly available on GitHub under the ttermish organisation.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Wedding photo app ditches cloud APIs to run AI upscaling privately on own server

Knipsmig, a QR-code-based photo sharing service for weddings and parties, has added a local 4x image upscaling feature to improve low-resolution guest uploads before they appear in printed photo books. The developer chose to run the Real-ESRGAN model (realesr-general-x4v3) entirely on-premise rather than use third-party APIs like Gemini or OpenAI, citing three key reasons: generative models alter faces rather than faithfully adding pixels, privacy regulations require updating data processing agreements whenever a new vendor handles guest photos, and per-image API costs become significant at scale. The chosen model is a compact 5 MB ONNX file with around 1.2 million parameters, running roughly 10 times faster on CPU than heavier alternatives. It was integrated into the existing Ruby on Rails editor pipeline via the onnxruntime gem, processing images in tiled 512-pixel segments with overlapping borders to avoid visible seams. Key optimisations included bypassing Ruby array marshalling for tensor data, instead writing raw bytes directly into OrtValue objects, which dramatically reduced overhead beyond the inference step itself.

0 comments Read more at DEV Community