A quality system that turns institute data into decisions you can verify.
Watch the 3-minute video (English and Arabic)
Every institute has numbers. Few have decisions. Quality Control System finds the problem, gives it an owner, and measures whether the fix worked.
Contents: The problem · The idea · A problem from start to finish · Product tour · Who uses it · How scores work · AI assistant · Trust and privacy · Programming skills · Status
Educational institutes collect plenty of data: enrollments, attendance, payments, assessments, leads, and classroom observations. It usually lives in separate files, and three things tend to go wrong:
- Reports arrive late. By the time someone compiles them, the moment to act has passed.
- Problems have no owner. Everyone can see a number, but nobody is responsible for changing it.
- Nobody checks the result. A decision gets made, and it's unclear whether it worked.
The system is built around one closed loop. Every feature serves one of its six steps.
| Step | What happens | Why it matters |
|---|---|---|
| 1. Data | Upload one Excel file. The system checks it and shows the result before you confirm. | Bad data is caught before it affects any number. |
| 2. Signal | 22 detection rules look for problems, with their size and records. | Problems surface without anyone having to search. |
| 3. Decision | Each decision gets one owner, a deadline, and a measurable target. | Responsibility is clear from the start. |
| 4. Action | Clear steps appear in each person's tasks. | Work reaches the people who do it. |
| 5. Verification | Someone who did not do the work confirms it was done. | The result doesn't rest on self-reporting. |
| 6. Impact | Before and after numbers show whether the decision worked. | The institute learns what actually works. |
Here is one example, using the test data:
- The signal appears. The system finds 602 enrollments with an unpaid balance, worth 696,920 AED.
- You check the evidence. A press opens the exact records behind the number.
- You decide. You choose a proposed action or write your own, then set an owner, a deadline, and a target. The AI assistant can suggest causes and options.
- The team acts. Steps appear in each person's tasks, and late work stands out.
- Someone else verifies. A different person confirms the work was done.
- The system measures. The new number is compared with the starting number and the target, and the result goes into the report.
All six departments on one screen, with a one-sentence reading of the current state and the work that needs you now.
Press any part of a chart, and the whole page filters by year, period, branch, course, status, teacher, or attendance level. The filters stay in the link, so a view is easy to share with a colleague who has access.
The dashboard follows the path from first contact to result, and adds a teacher performance map, courses and levels, enrollment seasons, and fees by branch.
A 3D map you can turn by dragging, with a column for each branch. The height can show learners, teachers, or quality. Pressing a column opens that branch's details.
Below the map, towers compare every department in every branch, and short sentences say where to start.
The full journey from lead to renewal, lead sources, reasons for losing leads, and a map of consultants.
Signals, evidence, proposed actions, and the full decision cycle in one place, with filters for status, importance, owner, branch, and deadline.
Observation history, the result of each item, a comparison with the team, the teacher's courses, strengths, and what to work on next. Classroom observations are filled in inside the system.
Filter by status, branch, course, level, teacher, consultant, year, balance, and date type, and open any record to trace it from sales to report.
For the whole institute, each branch, or each department, monthly or on demand. Each one includes a cover, summary, priorities, department results, proposed actions, and a sign-off area. They export to PDF and Excel, and issued reports never change.
| Role | What they get |
|---|---|
| Owner | The whole picture, final approval of decisions, and control of the AI assistant |
| Quality manager | Signals, decisions, reports, targets, checks, and users |
| Branch manager | Their own branch only |
| Academic manager | Teachers, observations, and academic results |
| Sales manager | Leads, consultants, and sales results |
| Teacher or action owner | Their own tasks and development record |
The six departments covered are Administration, Academic managers, Teachers, Sales, Students, and Marketing.
Every department has its own indicators, 34 in total. Each one is compared with a target that management agrees on, and weighted by its importance. The system also protects the score from weak data:
- An indicator without an agreed target isn't counted.
- A result based on too few cases is marked as a small sample instead of being judged.
- Each report states how many indicators its score rests on.
- Results that rely on estimated dates say so.
Management can change targets and importance from inside the system.
Ai is the on-screen assistant, a character made of particles that form the letter i. It drafts. You decide.
For reports, it drafts an executive summary, highlights, risks, and recommendations. A person edits and approves them before they appear.
For decisions, it suggests likely causes with a way to check each one, and gives options with honest pros and cons, drawing on past decisions about the same problem.
These two screens show illustrative examples on test data.
Three safeguards apply:
- Human review: nothing reaches a report or a decision without a person.
- Number checks: any number in a draft that isn't in the data is flagged for review.
- Privacy: names, phones, and emails of learners and staff never leave the system.
- Access by role and branch: each person sees only what they should, and each institute's data stays separate.
- Audit log: every change is recorded and can't be edited.
- Staff who leave: their account stops, they drop out of current indicators, and their history stays.
- Arabic and English: the complete interface in both, with right-to-left support.
- Any screen: desktop and phone, light and dark.
- Tested: 250+ automated tests cover permissions, data separation, and the main workflows.
| Area | Skills used |
|---|---|
| Backend | Python, Django, data modelling, role-based permissions, data separation between institutes |
| Data | Excel import with validation, calculating indicators, reports in PDF and Excel |
| Frontend | JavaScript, HTML, CSS, responsive design, right-to-left layouts, print design |
| Data visualization | Apache ECharts, cross-filtering charts, Canvas particle animation, 3D-style SVG maps |
| AI integration | LLM integration, structured outputs, removing personal data before sending, human review flows |
| Quality | Automated testing with pytest, audit logging, translation into Arabic and English |
All screens and numbers here come from test data. The system hasn't run on real institute data yet, so no real-world results are claimed. The AI assistant is in final testing.
بالعربي: الشرح الكامل
نظام جودة يحوّل بيانات المعهد إلى قرارات يمكن التحقق منها.
تجمع المعاهد بيانات كثيرة في ملفات منفصلة. فالتقارير تصل متأخرة، والمشكلات تبقى بلا مسؤول، ولا يعرف أحد هل نجح القرار.
- البيانات: ترفع ملف Excel، فيفحصه النظام قبل أن تعتمده.
- الإشارة: تكشف 22 قاعدة المشكلات وحجمها وسجلاتها.
- القرار: لكل قرار مسؤول واحد وموعد وهدف يمكن قياسه.
- الإجراء: تظهر الخطوات في مهام كل شخص.
- التحقق: يؤكد العمل شخص لم ينفذه.
- الأثر: تبين الأرقام قبل وبعد هل نجح القرار.
- مركز الجودة: الأقسام الستة في شاشة واحدة، وبماذا تبدأ.
- النظرة العامة: اضغط أي جزء في أي رسم لتصفية الصفحة كلها.
- الفروع: خريطة ثلاثية الأبعاد للإمارات، وأبراج تقارن الأقسام والفروع.
- المبيعات: من المهتم حتى التجديد.
- القرارات: الإشارات والأدلة ودورة القرار كاملة.
- المدرسون: صفحة تطوير لكل مدرس، والملاحظة الصفية داخل النظام.
- التقارير: للمعهد أو لكل فرع وقسم، بصيغة PDF وExcel، والمُصدر منها لا يتغير.
شخصية من جزيئات تشكل حرف i، يكتب المسودة وأنت تقرر. لا يظهر شيء قبل مراجعة إنسان، ويُنبّه إلى أي رقم غير موجود في البيانات، ولا تخرج الأسماء والهواتف والبريد من النظام أبدًا.
صلاحيات حسب الدور والفرع، وسجل تدقيق لكل تغيير، وواجهة كاملة بالعربية والإنجليزية، على الكمبيوتر والموبايل.
Python وDjango، وتصميم البيانات والصلاحيات، وJavaScript وHTML وCSS، والرسوم التفاعلية، ودمج الذكاء الاصطناعي، والاختبارات الآلية بـ pytest، ودعم العربية والإنجليزية.
كل الشاشات والأرقام هنا من بيانات تجريبية، ولم يعمل النظام بعد على بيانات حقيقية. والمساعد في مرحلة الاختبار الأخيرة.
About the source code: the code for this system is private. This page explains what it does and how it works, without internal details. To talk about the work, contact me through my GitHub profile.