⚠️ ATTENTION (PolyU Only)
This platform is open to PolyU students and staff.
Please connect to PolyUWLAN first, then access:
https://158.132.255.205:8080/login#/
Job interview preparation is often high‑stress, fragmented, and ineffective. 93% of candidates experience anxiety before interviews. Compounded by a lack of constructive feedback, many job seekers are left without a clear path to improvement.
PolyInterview is an end‑to‑end AI interview platform that bridges this gap. It delivers instant, structured, and multimodal feedback through an immersive real‑time Digital Human Interviewer.
We address the limitations of existing solutions:
- Human Coaching: costly ($100+/hr) and requires scheduling.
- Peer Mock Platforms: inconsistent quality and unreliable partners.
- Text/Voice AI: lacks visual immersion and “face‑to‑face” pressure.
Why now? The demand for scalable, realistic remote interview practice is surging. Only a real‑time digital human can provide the necessary immersive presence to truly prepare candidates for high‑stakes interactions, all while being affordable and available 24/7.
- 🤖 Interview Simulation: generate personalized question sets tailored to the company and role.
- 🗣️ Immersive Digital Human: experience real‑time lip‑sync interaction with a lifelike digital interviewer (powered by LiveTalking + Wav2Lip).
- 📝 Actionable Reports: get detailed, per‑question scoring and specific improvement guidance.
- 🔄 Closed‑Loop Flow: seamlessly move from question generation → interview practice → evaluation → iteration.
PolyInterview is a comprehensive, AI‑powered job interview simulation platform. It goes beyond simple Q&A by leveraging a sophisticated multi‑agent LLM architecture to guide candidates through the entire preparation journey—from interview setup to final performance evaluation.
Our goal is to help candidates master both general interview techniques and position‑specific skills through personalized, multimodal feedback in a highly realistic simulated environment.
The platform operates on an intelligent multi‑agent architecture that orchestrates the entire preparation journey:
- Question Generation Agents
- Interview Planner Agent: builds the interview plan and question list.
- Interviewer Agent: delivers questions and triggers follow‑ups.
- Performance Evaluation Agent: performs Feature Assessment (Professional Performance, Way of Expression, Non‑verbal Behavior, Oral Expression), then aggregates results into Aspect Assessment and Proficiency Assessment, producing evaluation results and improvement suggestions.
The platform ensures practice relevance by dynamically generating interview questions that align inquiry types with KSA/STAR/Big‑Five dimensions, tailored to the specific job description (JD).
The system strategically deploys different question types:
- Self‑introduction: to establish a baseline for communication traits.
- Behavioral Questions: to probe past actions, assessing traits and cultural fit.
- Skill QA: to directly assess technical capabilities and domain knowledge.
- Scenario Simulation: to evaluate problem‑solving abilities in hypothetical job situations.
Example: for a technical role requiring programming expertise and collaboration, the system generates Skill QA questions targeting technical requirements and Behavioral questions targeting team‑oriented traits. Mapping follows the diagram: KSA emphasizes Skill QA, STAR structures scenario/behavioral responses, and Big‑Five supports trait‑focused behavioral analysis.
Adaptive follow‑ups: To maintain a smooth and natural interview process, the system will check Consistency, Clarity, and Relevance of the answers, and adaptively generate follow‑up questions to challenge contradictory answers, probe unclear points, and handle irrelevant responses.
To provide objective and structured feedback, the assessment engine is grounded in robust theoretical frameworks that define candidate competency and communication effectiveness. KSA defines what is evaluated, STAR defines how responses are structured, and Big-Five provides a behavioral interpretation layer.
KSA Competency Framework (Knowledge, Skills, Abilities) [4]
- Knowledge (K): domain‑specific information applied to job tasks.
- Skills (S): learned proficiencies demonstrated through performance.
- Abilities (A): underlying capacity to perform observable behaviors.
STAR Method (Expression Technique) [5]
- Situation (S): the context or background of the scenario.
- Task (T): your responsibility or goal.
- Action (A): the specific steps you took.
- Result (R): the outcome and impact.
Big‑Five Personality Model (Behavioral Traits) [6]
- Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism as a trait lens for behavioral and communication assessment.
🟦 Multimodal Interview Assessment Framework
This section explains the internal evaluation architecture and theoretical grounding of PolyInterview. [1–3]
- Inputs: Text, Audio, Video from the interview session.
- Models:
- LLM for text understanding
- VLM for visual/non‑verbal cues
- Audio‑LLM for speech and vocal delivery
- Aggregation: scores flow from Feature → Aspect → Proficiency, where features are per‑response signals, aspects summarize cross‑response dimensions, and proficiency provides the overall evaluation.
- Hollandsworth Jr, J. G., Kazelskis, R., Stevens, J., & Dressel, M. E. (1979). Relative contributions of verbal, articulative, and nonverbal communication to employment decisions in the job interview setting. Personnel Psychology, 32(2), 359–367.
- Martín‑Raugh, M. P., Leong, C. W., Roohr, K. C., & Chen, X. (2023). Perceived conscientiousness and openness to experience mediate the relationship between vocal and visual features and hiring decision in the interview. Computers in Human Behavior Reports, 10, 100272.
- Martín‑Raugh, M. P., Kell, H. J., Randall, J. G., Anguiano‑Carrasco, C., & Banfi, J. T. (2023). Speaking without words: A meta‑analysis of over 70 years of research on the power of nonverbal cues in job interviews. Journal of Organizational Behavior, 44(1), 132–156.
- U.S. Office of Personnel Management. General Schedule Qualification Policies (KSA definitions).
- Indiana University School of Public Health Career Services. (2018). What is the STAR Interview Method?
- McCrae, R. R., & John, O. P. (1992). An introduction to the five‑factor model and its applications. Journal of Personality, 60(2), 175–215.
- 2026-01-31 — Added high‑demand queue stage for interview sessions and improved session orchestration.
- 2026-01-30 — Added user‑facing workflow manual, updated architecture visuals, and refreshed hero assets.
- 2025-12-15 — Introduced real‑time digital human interviewer powered by LiveTalking + Wav2Lip.
- 2025-11-01 — Launched structured assessment reports.