Frequently Asked Questions
Everything you need to know about deploying autonomous AI interviews at scale.
Q. How is the AI interviewer trained?
Our models are fine-tuned on millions of data points from elite technical and behavioral interviews. We utilize a mixture-of-experts (MoE) architecture to ensure domain-specific accuracy (e.g., a specialized sub-model just for assessing React knowledge).
Q. What happens if a candidate loses internet connection?
The session is instantly paused. The platform retains the precise state of the code editor and conversation. Upon reconnection, the candidate resumes exactly where they left off without penalty.
Q. Can candidates use external IDEs?
No. Our built-in code editor provides full IntelliSense and syntax highlighting. Forcing execution within our sandbox ensures absolute telemetry integrity.
Q. How do you detect AI-generated code from candidates?
Our Anti-Cheat engine analyzes keystroke dynamics, paste buffers, and compilation frequency. If a candidate pastes a perfect solution, the AI immediately pivots to ask deep, granular questions about specific lines of that code to verify understanding.
Q. Is the platform SOC2 Type II compliant?
Yes, we undergo annual SOC2 Type II audits. All data is encrypted at rest and in transit.
Q. How does the pricing scale?
We offer usage-based pricing per completed interview, with volume discounts for enterprise tiers conducting 10,000+ interviews annually.
Q. Does the AI suffer from hallucination?
We mitigate hallucination via a strict RAG pipeline hooked into verified documentation, and deterministic guardrails. The AI is constrained to evaluating against a predefined, immutable rubric.
Q. Can the AI understand heavy accents?
Yes. Our ASR (Automatic Speech Recognition) models are specifically trained on diverse global accents to ensure absolute fairness and zero voice-bias.
Q. How long does a typical interview last?
Standard configurations range from 45 to 60 minutes, though RecruiterOS allows custom durations down to 15-minute screening formats.
Q. What integrations do you support?
Native integrations for Greenhouse, Lever, Workday, and Ashby. Custom integrations are supported via our Webhook and REST API.
Q. Can we customize the AI's appearance or voice?
Yes. Enterprise clients can configure the vocal tone, pace, and visual representation to align with their employer brand.
Q. Is the code execution environment secure?
Extremely. Code runs in ephemeral Firecracker microVMs without external internet access, mitigating any risk of malicious code execution on our infrastructure.
Q. How do you ensure zero-bias in scoring?
The AI evaluates strictly based on the transcribed text, code logic, and predefined rubrics. It is blind to candidate demographics, age, or visual appearance.
Q. What happens if the AI encounters an edge-case candidate answer?
If confidence scores drop below a threshold, the system flags the session for human review, pausing automated final judgments.
Q. Can candidates request accommodations?
Yes. The platform supports extended time modifiers and text-based-only interfaces for accessibility compliance (WCAG 2.1 AA).
Q. Do you retain candidate video feeds?
By default, video is retained for 30 days to allow recruiter review via the Replay Engine, then permanently purged. Data retention policies are fully configurable.
Q. Can I test the AI interviewer before deploying it?
Yes, RecruiterOS includes a 'Sandbox Mode' allowing hiring managers to take the interview exactly as a candidate would.
Q. How do you prevent candidates from leaking questions?
Questions are dynamically generated based on a root concept. No two candidates ever receive the exact same prompt, rendering 'leak sites' ineffective.
Q. What programming languages are supported?
We currently support over 40 languages, including Python, Go, Rust, Java, C++, JavaScript/TypeScript, and domain-specific languages like SQL.
Q. What is the typical ROI for enterprise customers?
Our clients typically see an 80% reduction in engineering hours spent interviewing, and a 40% improvement in candidate-to-offer conversion rates due to objective screening.