IT Interview Questions and Answers for fresher's

Synchronous code executes sequentially line-by-line, blocking subsequent operations until the current one finishes. Asynchronous code (like setTimeout, Fetch, database queries) is delegated to browser Web APIs or Node worker threads, allowing the main single thread to continue without blocking. When completed, callbacks are queued into the Microtask/Task queue and pushed back to the call stack by the Event Loop.

The Virtual DOM is a lightweight in-memory JavaScript representation of the real DOM tree. When component state changes, React creates a new Virtual DOM tree, performs a 'diffing' algorithm against the previous snapshot (Reconciliation), and batches only the precise DOM node modifications needed to update the browser screen. This minimizes expensive real DOM reflows and repaints.

SQL databases (PostgreSQL, MySQL) are relational, table-based, enforce strict schemas, and prioritize ACID compliance, making them ideal for structured data with relationships (e.g. banking, e-commerce). NoSQL databases (MongoDB, Redis) are document/key-value based, schemaless or dynamic, and scale horizontally with ease, making them great for high-velocity unstructured data or caching.

REST (Representational State Transfer) is an architectural style based on stateless client-server communication, cacheability, and standardized HTTP methods (GET, POST, PUT, DELETE). Key status codes: 200 (OK), 201 (Created), 204 (No Content), 400 (Bad Request), 401 (Unauthorized), 403 (Forbidden), 404 (Not Found), 500 (Internal Server Error).

The CSS Box Model consists of content, padding, border, and margin. Under the default `content-box`, setting `width: 300px` applies only to content; padding and borders are added on top, expanding overall element size. With `box-sizing: border-box`, padding and borders are included within the specified width and height, preventing accidental layout breaks.

Supervised learning trains on labeled data (input-output pairs) to predict labels (Classification/Regression). Unsupervised learning finds intrinsic structures or clusters in unlabeled data (K-Means, PCA). Reinforcement learning trains agents to maximize cumulative rewards through trial-and-error environmental interactions.

Bias is error introduced by overly simplistic model assumptions (causing Underfitting—poor training and test accuracy). Variance is error introduced by extreme sensitivity to training data fluctuations (causing Overfitting—memorizing noise with high training but poor test accuracy). The goal is to find the sweet spot of minimum total error.

Precision = True Positives / (True Positives + False Positives)—how many predicted positive cases were actually positive. Recall = True Positives / (True Positives + False Negatives)—how many actual positive cases were captured. Recall is critical when missing a positive has severe consequences (e.g. cancer detection, fraud detection).

On-Page SEO involves optimizing elements directly on your own website (content quality, keyword targeting, title tags, internal linking, site speed). Off-Page SEO focuses on external signals to build domain trust and authority (backlinks from authoritative sites, brand mentions, social signals, digital PR).

Ad Rank determines your ad's placement on the search results page. It is calculated by: Ad Rank = Maximum CPC Bid × Quality Score × Expected Impact of Ad Assets (extensions) and format. Quality score depends on expected click-through rate (CTR), ad relevance, and landing page experience.

ROAS (Return on Ad Spend) measures gross revenue generated per dollar spent strictly on advertising (Revenue / Ad Spend). ROI (Return on Investment) accounts for total costs including ad spend, cost of goods sold, overhead, and salaries ((Net Profit / Total Costs) × 100).

WHERE filters individual rows BEFORE aggregations occur and cannot be used with aggregate functions (like SUM, AVG). HAVING filters grouped records AFTER the GROUP BY aggregation has been calculated.

A Star Schema consists of one centralized Fact table containing quantitative business metrics (e.g. sales, revenue) connected directly to multiple surrounding Dimension tables containing descriptive context (e.g. customers, dates, products). It simplifies analytical queries, reduces JOIN complexity, and speeds up BI performance.

Outliers can be detected using statistical methods like Z-scores (values > 3 standard deviations from mean) or IQR (Interquartile Range: < Q1 - 1.5*IQR or > Q3 + 1.5*IQR), and visualized using boxplots. Depending on context, outliers can be removed, capped (winsorization), transformed (logarithmic), or kept if they represent genuine anomalies.