Featured Engineering Case Studies

Problem: University internship placement and student monitoring across Francophone Africa still rely on paper forms, scattered emails, and manual spreadsheets, leading to roughly 30% unverified agreements and zero actionable data.

Solution: A complete cloud infrastructure connecting 5 core roles (universities, corporate hosts, students, academic supervisors, and ministries) with automated approval workflows, tamper-proof digital agreements, and a real-time employability observatory.

Outcome: Live operational MVP on stageconnect.app. 2026-2027 academic pilot in preparation, targeting 10 to 15 private universities in Benin (signatures from October 2026).

Problem: Every year, thousands of Beninese high school graduates select university majors by default or misinformation, risking dropout due to misalignment with their actual Baccalaureate grades and cognitive affinities.

Solution: An AI-powered academic advisor combining Baccalaureate score analysis, RIASEC psychometric profiling, real-time official MESRS ranking algorithms, and an illustrated Parent Report exportable as images, accessible on web and WhatsApp.

Outcome: Launched in July 2026: 184 landing page visitors, 80 reached the payment page and 7 became paying customers. Usage data now shapes the V2 architecture.

Problem: Private and public schools struggle with fragmented legacy software, brittle spreadsheets, and severe communication latency with parents regarding student attendance and tuition status.

Solution: A multi-school management platform for Benin: 4 apps, one per role (super admin, principal, teacher, class coordinator), covering per-class attendance, QR code check-in, grades and report cards, student follow-up, payments and reminders, and parent updates over WhatsApp or SMS.

Outcome: Built in 3 months, with strict per-school data isolation (PostgreSQL Row-Level Security) validated by concurrency and load tests before going to production.

Problem: A South African tour operator needed a modern web platform delivering fast page loads on international mobile connections alongside high organic search discoverability and intuitive bookings.

Solution: A booking platform for tours and accommodation across Africa: public website in 5 languages, destinations by country, booking flow with payment and PDF travel voucher, customer area and blog.

Outcome: Complete frontend shipped to production between May and July 2026 across 15 reviewed and merged contributions, with layouts adapted to mobile and large screens alike.

Problem: Personal budgeting tools impose excessive manual friction and overlook cash-heavy, informal multi-wallet habits across West Africa, causing users to abandon spreadsheets within days.

Solution: A voice-native mobile application powered by a proprietary AI correction loop: users speak out their day, AI extracts multi-wallet expenses (Cash, MTN MoMo, Wave) and aligns theoretical balances with physical money without shame or complex math.

Outcome: Landing page and waitlist live on budgetvox-app.sinda.pro; the mobile app is in final internal testing before the beta opens: Deepgram Nova-3 voice transcription (< 300 ms latency) and structured AI expense extraction.

Problem: A financial education academy needed a platform to sell its online courses (stock market, BRVM) and publish its analyses without relying on social media.

Solution: A complete e-learning platform: instructor area (courses, videos, resources, analytics), learner area (catalog, mobile money payments in FCFA, video player, progress tracking, quizzes), an analysis blog and email or Google sign-in.

Outcome: Live on safre-academie.com, with more than 250 registered users.

Problem: In Benin, filings with the APDP (the personal data protection authority) are still assembled by hand: scattered documents, re-typing, omissions and long correction loops. And the platform that would centralise these files would itself become a sensitive target.

Solution: A platform that digitises the whole cycle: guided filing, detection of the applicable regime (declaration or authorisation), documents sealed with SHA-256, then a technical pre-audit of the data controller's website (OWASP ZAP, OSV) whose findings an AI correlates and explains. The AI suggests, a human decides: every finding awaits validation and the examiner alone makes the decision.

Outcome: Security foundation delivered and tested: access control by explicit permission, tenant isolation through PostgreSQL Row-Level Security, httpOnly cookie sessions with secret rotation, audit log. 271 automated tests pass; the pre-audit and examination workflow are in development.

Problem: External attack surface management (EASM) tools produce raw results: the same server counted once per DNS name, a visible exposure treated as a confirmed vulnerability, and prioritisation that is opaque or missing.

Solution: A Python command-line tool that cross-checks passive sources (DNS, Certificate Transparency, RDAP), groups names that point to the same server, and gives each observation one of 4 confidence levels before ranking it with explainable criteria. Reports are written in French. Active testing is limited to a lab and to domains with written authorisation.

Outcome: Live demo with fictitious data (reserved domains, documentation IP addresses and AS number), published method and roadmap. An exploratory, passive-only study compares the DNS visibility of government domains in 7 countries. The command-line prototype is in development.

Problem: HackerLab is Benin's national cybersecurity competition, run by ASIN. The 2026 scenario, "Operation Tervalis": defend a fictional nation against masked actors by identifying the attackers, tracing attack chains and protecting critical infrastructure.

Solution: A team of 4 organised as a cell: one role per challenge family (pwn and reverse, AI and attribution, web and OSINT, forensics and crypto), a shared repository with a per-challenge template (notes, files, validation), a tool installation script and a MITRE ATT&CK guide for attribution. Before the competition: training on Root-Me.

Outcome: Final ranking: ByteMeIfYouCan 11th with 13,319 points, the maximum score also reached by the top ten teams, separated by time. On my own: 93 flags, 10,479 points, 5th in the individual ranking and 4 first bloods (first in the whole competition to solve Bohemian Rhapsody, The Ashen Seal, Loose Cage and Phone Home), 8th on the BloodBoard. On 29 August, my points made up 83% of the team score.