Christian Asnel Ngoulla

• • Architected and shipped a full-stack MLOps platform (Next.js / FastAPI / PostgreSQL) from 0→ 1, enabling end to-end dataset management, model training, evaluation, and deployment workflows — reducing ML iteration cycle time by ~40%.

• • Designed an IDE-like visual ML pipeline editor integrated with Kubeflow and GitLab CI, eliminating manual pipeline configuration steps and cutting data-scientist onboarding time by ~60%.

• • Led architecture migration to micro-frontends (Module Federation) + microservices across 3 teams, improving independent deployment frequency by 3× and reducing cross-team merge conflicts by ~70%.

• • Owned end-to-end production delivery on Kubernetes (Helm/Kustomize) + GitLab CI/CD; introduced automated health checks and rollback strategies, achieving 99.9% deployment success rate over 18 months.

• • Built and integrated a real-time collaborative file management service (WebSocket), enabling live code sharing across AI workflow teams with sub-100ms sync latency.

• • Designed and delivered an image-labeling application (Vue.js / FastAPI / WebSocket / PostgreSQL) used by annotators across 3 product lines, processing 500K+ images to date.

Stack: Kubernetes

◦ · Kubeflow

◦ · Helm/Kustomize

◦ · GitLab CI/CD

◦ · Docker

◦ · FastAPI

◦ · TypeScript

◦ · Next.js/Vue

◦ · PostgreSQL

◦ · Azure

◦ · GraphQL

◦ · WebSocket

• • Delivered Pro:Tec Sports — a production Angular + Java web platform for pension claim enforcement — from spec to launch, applying TDD with >80% test coverage and deploying to AWS with zero-downtime blue/green strategy.

• • Built serverless event-driven architecture for Radien (Next.js + AWS Lambda + SQS/SNS + ECS Fargate), cutting infrastructure costs by ~35% vs. previous always-on approach and handling 10K+ daily events.

• • Defined REST APIs and relational data models; wrote comprehensive integration tests; shipped all features on schedule across cross-functional squads of 6+ engineers.

• • Mentored 3 full-stack engineers; established code review standards and architectural decision records (ADRs) that were adopted team-wide, reducing PR cycle time by 25%.

Stack: Java (Spring/Hibernate)

◦ · Angular

◦ · Next.js

◦ · AWS Lambda

◦ · SQS/SNS

◦ · ECS Fargate

◦ · Keycloak

◦ · PostgreSQL

• • Built the visitor-facing web app for an AI-powered art recognition product deployed across multiple museum clients, maintaining 99.5% uptime in production.

• • Hardened Azure Pipelines quality gates for Azure Functions + PostgreSQL releases, automating validation checks and eliminating risky manual deployment steps — reducing deployment incidents by 80%.

◦ · Azure Functions

◦ · Azure Pipelines

◦ · Python

◦ · PostgreSQL