1PGTINDO

Stage 1

Story mode

Seven years of software, played in order. Each job is a stage, each thing shipped a boss defeated.

  1. Training 1

    Prépa Vogt

    Preparatory school — Mathematics & Physics (MPSI/MP)

    2 yearsCameroon

    The training stage: two years of maths and physics before choosing a weapon.

  2. Training 2

    Institut Universitaire de la Côte · 3iL

    Engineering degree, Computer Science — Security & Web Development

    3 yearsCameroon

    The 3iL Ingénieur curriculum: where security and the web became the main build.

  3. Stage 1

    Qognitiv Technology

    Software Engineer

    Aug 2019 – May 2020Cameroon

    Backend and machine learning engineer on an active liveness detection stack for embedded banking use cases.

    • Boss defeated: LivenessBuilt a head detection model and an eye-blinking detection pipeline.
    • Boss defeated: PaperworkBuilt OCR for document ID processing.
    • Boss defeated: The backendWrote the service code supporting the liveness detection flow.

  4. Stage 2

    Workerly (now Workbud)

    Software Engineer

    Dec 2020 – Aug 2021Remote from Cameroon, for a New York–based company

    Backend contractor on the company’s HR tool, before its pivot to a workforce collaboration platform.

    • Boss defeated: The APIBuilt the API of the HR tool with Django, Django REST Framework and PostgreSQL.
    • Python
    • Django
    • Django REST Framework
    • PostgreSQL

  5. Stage 3In progress

    Unissey

    Software Engineer — technical referent, API/SDK team

    Sep 2021 – PresentParis

    Security-critical products in identity verification and biometrics, from front-end SDK integrations to distributed backend services.

    • Boss defeated: CaptureDeveloped web SDKs (web components with React and Angular wrappers) for the video, selfie and document ID capture flows used in identity verification.
    • Boss defeated: 20M+ a yearBuilt and maintained the customer-facing APIs orchestrating biometric analysis pipelines — presentation attack detection, deepfake detection, injection attack detection, face comparison, age verification — deployed on-premise on customer infrastructure and as SaaS, handling 20M+ identity verifications per year.
    • Boss defeated: Legacy DebtLed a major refactor of the codebase to address legacy debt and high team turnover; the team has been stable and shipping faster ever since.
    • Boss defeated: The Blind SpotIntroduced OpenTelemetry and distributed tracing across backends, including context propagation via baggage, to correlate customer on-prem apps, SaaS and internal load-test dashboards.
    • Boss defeated: The GatekeeperDeveloped a suite of Keycloak extensions (custom authenticators, JPA entities, endpoints, themes, required actions) enabling authentication via facial biometry with the team’s video capture SDK.
    • Boss defeated: LabelsContributed to an internal data labeling platform letting the R&D team inspect GDPR-consented media, flag and annotate sessions, and transfer data for model training, testing and evaluation.
    • Boss defeated: Manual ModeDrove the team’s migration from manual coding to agentic development: ran workshops on agent harnesses, spec-driven development and skills, and adapted the API/SDK codebases to be agent-friendly.
    • TypeScript
    • Node.js
    • React
    • Java (Keycloak)
    • Rust (WebAssembly)
    • C++ (N-API)

Move list