How MockExperts Screens Engineering Candidates End-to-End — A Complete B2B Hiring Flow Demo (2026)

Every assessment on MockExperts begins with a JD (Job Description) and optionally a stack of candidate résumés. Before you generate a single question, the platform's AI résumé screener ingests the uploaded PDF résumés and produces a structured signal for each candidate:

• JD Match Score: How well does the candidate's stated experience align with the required skills, years of experience, and seniority level in the JD?
• Skill Gap Summary: Which listed requirements are missing, partially met, or over-qualified for?
• Seniority Signal: Is this candidate a genuine Senior engineer, or is the résumé papering over a 2-year experience gap?
• Red Flags: Unexplained gaps, vague project descriptions, or mismatched tech stacks flagged in plain English.

This means your recruiter doesn't spend 45 minutes reading 30 résumés. They open a ranked, pre-screened shortlist and move directly to assessment creation with high-signal candidates already prioritized. The screening model is calibrated against real MAANG hiring rubrics, not generic keyword matching.

Why This Matters for Non-Technical Recruiters

Résumé screening in most ATS tools is a keyword filter. MockExperts' AI reads résumés the way a senior engineering manager does — contextually. It understands that "5 years of React experience" on a résumé dominated by jQuery plugins is a very different signal from the same phrase on a résumé with open-source contributions and production-scale architecture write-ups.

Step 2: Assessment Creation — Rubric Evaluation, Not Just Questions

Once you have a candidate shortlist, you create an Assessment — the interview blueprint for a specific role. This is where MockExperts diverges entirely from platforms that just spit out generic Leetcode problems.

Customizing the Evaluation Rubric

Every assessment starts with a configurable rubric: the weighted evaluation framework that tells the AI exactly what kind of engineer this role demands. You set weights across five dimensions:

Dimension	What It Evaluates	Example Weight (Backend Eng)
DSA	Algorithmic thinking, time/space complexity, problem decomposition	30%
System Design	Distributed systems, scalability patterns, trade-off articulation	35%
Code Quality	Clean code principles, naming, modularity, error handling, testability	20%
Behavioral	Ownership, conflict resolution, cross-functional communication, STAR fluency	15%
Domain-Specific	Optional: ML, security, mobile, frontend performance, etc.	Configurable

The AI uses these weights when generating questions and when scoring candidate responses — so every score is rubric-anchored, not impressionistic. A candidate who aces behavioral but blanks on system design doesn't get a misleadingly high overall number.

CRUD Operations on Your Question Bank

After generating the question set, you get full editorial control via a complete CRUD interface:

• Add your own custom questions (useful when a hiring manager has role-specific pet questions)
• Edit any generated question — tighten the scope, change the context, adjust difficulty
• Delete questions that don't fit the role's actual day-to-day requirements
• Reorder questions to control the interview pacing — easier questions first to settle nervous candidates

Regenerate on Comment — The Most Powerful Recruiter Feature

This is the feature that most impresses hiring managers when they first see it. On any question card, you can type a comment in plain English — no technical vocabulary required — and the AI will regenerate only that question according to your instruction. Examples:

• "Make this system design question more focused on database sharding, not caching."
• "This DSA question is too easy for a senior. Make it harder."
• "The behavioral question is too generic — make it specific to a startup context where the team has no QA."
• "Change the code quality scenario from Python to Go."

The AI reads your comment, preserves the rubric intent, and produces a new question that incorporates your feedback exactly. You can iterate as many times as needed. You can also regenerate the entire question set in one click if you decide the whole batch needs a fresh pass — useful when you change the role seniority mid-process or receive updated JD requirements from the hiring manager.

Step 3: Approve, Publish & Send Branded Candidate Emails

When the question bank is finalized and the rubric is locked, the recruiter hits Approve & Publish. The assessment goes live and is now ready to be dispatched to candidates.

Customizable Email Templates

MockExperts does not send generic system emails. Every candidate communication is customizable through a branded email template builder. You control:

• Company name and logo — candidates see your brand, not MockExperts
• Opening message — personalize the tone (formal corporate vs. casual startup)
• Role details — the assessment title, expected duration, and instructions are auto-populated
• Deadline — set a hard expiry window for the assessment link
• Support contact — route candidate queries to your HR team's inbox, not ours

The outgoing email contains a unique, candidate-specific assessment link tied to their email address. The link is non-transferable — it can only be authenticated with the exact email it was sent to. No link sharing, no proxy attempts.

What the Candidate Receives

A clean, professional email lands in the candidate's inbox. One button click. No password to create. No app to download. They land on the exam setup page, grant camera and microphone permissions, and they're in.

Step 4: The Candidate Experience — Multi-Modal, Proctored, Speaking-First

This is where MockExperts is categorically different from every other technical assessment platform. Candidates do not write code and run it. They think out loud, speak their logic, and explain their reasoning — exactly like in a real MAANG technical loop.

Login With the Referred Email

The candidate clicks their unique link from the invitation email. The platform validates that the authenticating email matches the invited address. Any attempt to log in with a different account is rejected immediately. This single gate eliminates the most common form of proxy cheating: having a stronger engineer take the assessment under a different identity.

The Proctored Interview Environment

Once authenticated, the candidate enters a live-proctored session with camera and microphone active. The platform monitors:

Signal	What Is Tracked	Why It Matters
Tab Switches	Every time the candidate leaves the exam window	Flags potential external resource use
Copy-Paste Attempts	Clipboard events within the exam environment	Detects pasting from external sources
Face Presence	Camera checks that the registered face is in frame	Detects face substitution or candidate absence
Voice Activity	Microphone monitors continuous speech or suspicious silences	Validates genuine live explanation
Time on Question	Per-question time spend logged granularly	Reveals confidence vs. struggle patterns

All of these signals are aggregated into an Integrity Score that appears on the final report — not just a number, but a timestamped event log so the recruiter can see exactly what happened and when.

Speaking Evaluations — DSA, System Design, Code Quality, Behavioral

For each question, the candidate speaks their answer aloud. The platform's real-time transcription engine converts their speech to text with high fidelity — technical terms, algorithm names, complexity notation, all handled correctly. The transcribed text is what the AI evaluates against the rubric.

This design is intentional and powerful for several reasons:

• DSA without a code runner: Candidates explain their approach — the algorithm choice, the time/space trade-off, the edge cases they'd handle, and why their solution is better than the naive approach. This is precisely how MAANG companies run their early DSA screens. The logic matters far more than the ability to type syntax correctly under pressure. No LeetCode tricks. Just reasoning.
• System Design by speaking: Candidates explain their architecture verbally — components, data flow, database choices, scalability considerations, failure modes. The AI evaluates whether they covered the core pillars and whether their trade-off articulation reflects the depth expected at the given seniority level.
• Code Quality without executing code: Candidates are shown a scenario or a code snippet and explain: How would they structure this? What would they name the modules? Where would they add error handling? What patterns are they applying and why? This evaluates whether they actually understand clean code principles versus just being able to pass test cases.
• Behavioral in STAR format: Candidates respond to situational questions verbally. The AI evaluates the completeness of their STAR (Situation, Task, Action, Result) framing, the specificity of their examples, and the quality of their self-reflection.

The speaking-first model also eliminates a major equity issue with traditional coding platforms: candidates who are strong engineers but slow typists (common in non-Western markets and across seniority levels) are no longer disadvantaged by a format that conflates typing speed with engineering quality.

Step 5: Detailed Feedback Reports — Designed for Non-Tech Recruiters

When a candidate completes their interview, the recruiter sees a report that is unlike anything produced by legacy screening tools. It is not a score out of 100 with a percentile bar. It is a structured, narrative evaluation that any hiring stakeholder — regardless of technical depth — can read, act on, and defend.

What the Report Contains

Every completed assessment generates a full report structured across six sections:

1. Executive Summary (Plain English)

A 3–5 sentence paragraph written in natural language, summarizing the candidate's overall performance. No jargon. Example:

"Priya demonstrated strong conceptual understanding of distributed systems — she correctly identified the need for a message queue and proposed Kafka with logical justification. Her DSA reasoning was solid on the binary search problem but she missed the edge case for an empty array. Code quality explanations showed familiarity with clean architecture principles. Behavioral responses were specific and well-framed. Overall integrity was clean — zero tab switches, zero copy-paste attempts."

2. Per-Dimension Scorecard

A breakdown of the candidate's performance against each rubric dimension, with the weighted contribution to the final score displayed clearly. Non-technical readers immediately see: where the candidate excelled and where they fell short — without having to interpret code.

3. Ideal Architecture vs. Candidate Response (Side-by-Side)

For system design and code quality questions, the report shows a side-by-side comparison:

• Left panel — Ideal Answer: What a Senior/Staff-level engineer at a top company would say, covering all mandatory components.
• Right panel — Candidate's Transcribed Response: What the candidate actually said, annotated with colour-coded markers (green = correct, amber = partially correct, red = missing or wrong).

This lets even a non-technical recruiter instantly visualize the gap between the expected bar and the candidate's actual performance — without understanding the underlying technology.

4. Specific Strengths & Gaps

The AI surfaces the top 3 strengths and top 3 gaps for each candidate in bullet-point form. These are specific — not "good at algorithms" but "correctly identified that a hash map reduces the lookup from O(n) to O(1) and explained why this matters at 10 million records per day scale."

5. Integrity Report

A timestamped event log of all proctoring signals: tab switches with duration, copy-paste attempts with context, face-detection losses, and microphone anomalies. The recruiter can see at a glance: Was this a clean interview? Or does the performance need to be discounted for integrity concerns?

6. Hire / No Hire Verdict

The most important section. MockExperts produces a clear, defensible verdict:

Verdict	What It Means	Typical Score Range
✅ Strong Hire	Candidate exceeded rubric expectations on majority dimensions. Move to offer discussion immediately.	85–100
🔵 Hire	Candidate met the bar on all critical dimensions. Proceed to final round or reference checks.	70–84
⚠️ Borderline	Mixed performance. Consider a follow-up targeted technical call on the weak dimension.	55–69
❌ No Hire	Candidate did not meet the minimum bar. Report explains exactly which dimensions failed and why.	0–54

Critically, the verdict is not just a number. It includes a 2–3 sentence rationale explaining why the system reached that verdict — grounded in the actual rubric performance, not opaque model scoring. This makes the verdict defensible in hiring committee discussions and protects the company from unconscious bias claims.

The Full Flow — From Résumé to Verdict in One Diagram

To summarise the complete B2B hiring pipeline:

1. Upload JD + Résumés → AI résumé screen ranks and flags candidates by JD match.
2. Create Assessment → Configure rubric weights, generate MAANG-style questions, iterate via comments.
3. Approve & Publish → Lock the question bank, customise the candidate email template, dispatch unique links.
4. Candidate Authenticates → Email-gated login. Camera and microphone activated. Proctoring begins.
5. Multi-Modal Interview → Candidate speaks their DSA logic, system design, code quality reasoning, and behavioral answers. Transcription runs in real-time.
6. AI Evaluation → Transcripts scored against the rubric. Integrity signals aggregated. Report generated.
7. Recruiter Reviews Verdict → Plain-English executive summary, side-by-side architecture comparison, and a clear Hire/No Hire verdict with rationale.

Why MockExperts Beats Traditional Screening Platforms

vs. HackerRank / Codility

HackerRank tests whether candidates can pass automated test cases — not whether they can engineer at scale. It rewards memorized LeetCode patterns and penalizes strong engineers who think out loud. MockExperts tests engineering judgment, communication, and reasoning — the actual signals that predict on-the-job performance.

vs. Manual Technical Interviews

A senior engineer's time costs $300–500/hour all-in when you factor in scheduling overhead and opportunity cost. Running 20 first-round interviews per open role is a $10,000+ spend before any useful signal. MockExperts asynchronously screens all 20 candidates in parallel, delivers ranked reports, and surfaces only the top 3–4 for a final human round.

vs. Generic AI Screening Tools

Most "AI screening" tools are keyword filters dressed up with machine learning branding. MockExperts generates role-specific, rubric-anchored questions, evaluates spoken reasoning rather than syntax, and produces verdict reports grounded in what the candidate actually said — not proxy metrics like résumé keyword density.

Frequently Asked Questions

Can non-technical recruiters set up assessments without engineering help?

Yes. The assessment creator is designed for non-technical users. The most complex decision a recruiter needs to make is the seniority level and the core skill focus (e.g., "Senior Backend Engineer, primary focus: System Design"). The AI handles rubric weighting, question generation, and evaluation logic entirely. No engineering configuration required.

How does the voice transcription handle technical jargon and algorithm names?

The transcription engine is fine-tuned on a corpus of engineering interview speech — it correctly handles terms like "Big O notation," "idempotency," "consistent hashing," "Kafka consumer group," and dozens of other technical concepts that generic speech-to-text models mangle. Candidates are instructed to speak clearly and at a natural pace; the system handles the rest.

What happens if a candidate's internet connection drops mid-interview?

The interview session is designed to be resumable. The system automatically saves progress at the end of each question. If a candidate drops and reconnects within the assessment window, they continue from where they left off. The proctoring log notes the disconnection but does not automatically flag it as an integrity issue unless it is repeated or accompanied by other signals.

How many candidates can we invite per assessment?

This depends on your subscription tier. The Pilot plan (free, 7-day trial) supports up to 100 candidates per month across 5 assessments. The Business plan scales to 1,000 candidates per month across 50 assessments. Enterprise plans have custom limits. See our pricing page for the full breakdown by region (USD and INR).

Can we white-label the candidate experience?

Partial white-labelling is available from the Business tier: the candidate email templates carry your company branding, and the assessment interface displays your company name and logo. Full white-label domain support (assessments served from your own subdomain) is available on Enterprise.

Is the Hire/No Hire verdict legally defensible?

MockExperts produces a structured, criteria-referenced evaluation — not a black-box score. The verdict is anchored to specific rubric dimensions with documented rationale, which is the gold standard for defensible hiring decisions. We recommend using MockExperts verdicts as one structured input in a multi-signal hiring process rather than as the sole determinant of hire/no hire.