Practise job interviews with an AI hiring manager — across 6 roles, each scored on its own rubric.
A single-file, zero-dependency web app for interview practice. The AI plays the hiring manager; you answer as the candidate. When you're done, the app scores your answers against the rubric that role is actually hired on and gives you targeted improvement tips.
Built for classroom and workshop use — it runs entirely in the browser with no backend,
no build step and no installation. Just open index.html.
- 6 job roles, each with its own interviewer persona, opening question and question bank
- Role-specific rubrics — 6 criteria × 5 marks = 30 per role (no generic scoring)
- Adaptive interviewing — vague answers get probed ("Can you give me a specific example?") before the interview moves on
- 3 difficulty levels — Entry (supportive), Mid (standard), Senior (challenges your answers)
- Scored feedback — per-criterion marks, an overall hiring signal, what worked well, 3 improvement tips and a model answer upgrade rewriting your weakest answer in STAR form
- Demo mode — a local scripted interviewer + keyword rubric scorer, so a full class can practise with no API key and no cost
- Bring your own key — OpenAI or MiniMax, entered at runtime and kept only for the browser tab
- Copy feedback / download transcript as a
.txtfor reflection or submission - Responsive, accessible (44px touch targets,
aria-livechat, keyboard-friendly), works offline in demo mode
| Role | Interviewer | Focus |
|---|---|---|
| Software Engineer | Priya Raman, Engineering Manager | System design trade-offs, debugging, code quality |
| Data Analyst | Marcus Tan, Head of Analytics | SQL reasoning, stakeholder handling, "so what?" |
| Product Manager | Elena Fischer, Director of Product | Prioritisation, metrics, stakeholder influence |
| Digital Marketing Executive | Daniel Ong, Marketing Director | Campaign ROI, channel strategy, testing |
| Customer Service Officer | Aisha Rahman, Customer Success Lead | Empathy, difficult customers, composure |
| HR Executive | Grace Lim, HR Business Partner | Confidentiality, ethics, difficult conversations |
Every role is scored on 6 criteria, 5 marks each (30 total). The criteria differ per role — a Software Engineer is graded on System design & trade-offs, while an HR Executive is graded on Confidentiality & ethics.
See all six rubrics
Software Engineer — Technical depth · Problem solving & debugging · System design & trade-offs · Code quality & testing · Collaboration & communication · Ownership & impact
Data Analyst — Analytical rigour · Technical tooling (SQL/Excel/BI) · Business acumen · Storytelling & visualisation · Stakeholder management · Attention to detail
Product Manager — Customer & problem framing · Prioritisation & trade-offs · Metrics & outcome thinking · Stakeholder influence · Execution & delivery · Strategic thinking
Digital Marketing Executive — Campaign strategy · Data & ROI literacy · Creative & content judgement · Channel & tool knowledge · Testing & iteration · Commercial impact
Customer Service Officer — Empathy & active listening · Problem resolution · Communication clarity · Composure under pressure · Product & process knowledge · Service mindset
HR Executive — People judgement · Confidentiality & ethics · Employment knowledge · Communication & difficult conversations · Process & organisation · Stakeholder partnering
Scoring rewards concrete examples, metrics, personal ownership ("I" not just "we") and honest reflection. It penalises generic claims ("hard worker", "team player"), rambling and answers with no result.
Option 1 — use the hosted version: alfredang.github.io/AIInterviewing
Option 2 — run locally:
git clone https://github.com/alfredang/AIInterviewing.git
cd AIInterviewing
open index.html # or: python3 -m http.server 8000There is no build step and no dependencies.
Demo mode needs no key. For a real AI interviewer:
- Choose OpenAI (GPT-4o-mini) or MiniMax (M2.7) in Interview mode
- Paste your key and click Save key
- Start the interview
| Provider | Endpoint | Model |
|---|---|---|
| OpenAI | api.openai.com/v1/chat/completions |
gpt-4o-mini |
| MiniMax | api.minimax.io/v1/chat/completions |
MiniMax-M2.7 |
Key handling: the key is stored only in
sessionStorage— it lives in that browser tab, is cleared when the tab closes, is never committed to this repo and is never sent anywhere except the provider you select. Calls go browser → provider directly; there is no backend. For classroom use, prefer a training key with a low spend limit.
- Pick a role and an interview level
- Read the rubric shown for that role so you know what is being assessed
- Answer each question using STAR — Situation, Task, Action, Result
- Aim for 6–10 questions, then click Get Interview Feedback
- Review your scores, tips and the model-answer rewrite — then restart and improve
Vanilla HTML + CSS + JavaScript in one self-contained index.html (~1,100 lines).
No frameworks, no build tooling, no external requests except the AI provider you choose.
AIInterviewing/
├── index.html # the entire app: markup, styles, roles, rubrics and logic
├── screenshot.png # README preview
└── README.md
Design system and interaction model adapted from the AI Coaching practice lab. Built with Claude Code.
Tertiary Infotech Academy Pte Ltd · UEN 201200696W 🌐 tertiaryinfotech.com
For training use only · All roles and interviewers are fictional
© 2026 Tertiary Infotech Academy Pte Ltd · UEN 201200696W