Data platforms · Cloud · AI
Yves De Boeck
Data & AI Engineer
Building production-grade data platforms, cloud infrastructure, and AI systems that solve real industrial problems.
About
Who I am
I'm a Belgian engineer with a deep passion for building systems that work in the real world - not just on paper. My career spans full-stack web development, IoT, cloud infrastructure, data engineering, and AI - and I'm most energised when those disciplines intersect.
I hold two master's degrees from the University of Antwerp, where I graduated magna cum laude. After being offered a PhD in AI for autonomous vehicles, I chose industry to work on concrete problems at scale.
Over the past six years I've worked across energy, healthcare, and industrial IoT - building everything from embedded device firmware to production ML pipelines. What keeps me interested is the point where those layers meet: getting data off real hardware, through a platform that holds up, and into something people actually use.
React, FastAPI, embedded C++ on Yocto - from device to cloud.
Databricks, PySpark, dbt, Delta Lake. Batch and streaming at scale.
RAG systems, LLM agents, MLOps with Azure AI Foundry and MLflow.
Experience
Where I've worked
Atlas Copco
Data & AI / Full Stack / Cloud Engineer
Jan 2024 – Present
Belgium
- ›Designed and built a cloud-native IoT fleet management platform on Azure, ingesting real-time telemetry from industrial compressors worldwide via MQTT and IoT Hub, streamed live to the frontend via WebSockets.
- ›Wrote the embedded C++ component on a custom Yocto OS for compressor controllers, bridging the internal MQTT bus with the Azure IoT Hub.
- ›Built and maintained production APIs consumed by field service technicians on customer sites.
- ›Set up a Databricks data platform from scratch: medallion architecture, PySpark, dbt, Delta Lake, Unity Catalog, DLT - ingesting SQL, JSON, XML, Kafka, and text sources in batch and streaming.
- ›Developed a RAG-based internal knowledge app using Azure AI Search and Prompt Flow, deployed across multiple departments with per-department document indexes.
- ›Created an agentic AI assistant (similar to OpenClaw) using Azure Agent Framework (Semantic Kernel) and Azure AI Foundry, integrated with Slack - handling ticket management, PR reviews, security assessments, and CI/pipeline status reports.
- ›Built an agentic security scanner published as an Azure DevOps task: auto-detects stack, runs scanners, triages findings, fixes code in an agentic loop, and auto-generates a PR for final fixes.
- ›Full IaC with Bicep and Terraform across all projects; CI/CD with Azure DevOps and Jenkins.
TrinetX
Software Engineer
Sep 2021 – Dec 2023
Ghent, Belgium
- ›Full-stack development on a large-scale healthcare analytics platform (Java/Spring + React) used by hospital networks worldwide.
- ›Built ETL services in Python processing healthcare data across Vertica, Snowflake, Redis, and MongoDB.
- ›Designed and built a new observability service for ETL processes from scratch, including architecture decisions, CI/CD setup, and AWS cloud deployment.
- ›Operated in a strict TDD, SOLID, domain-driven design environment with Jenkins and GitLab CI.
Essent
AM Software Engineer – Billing Processes
2020 – Sep 2021
Belgium
- ›Analysed and resolved issues across Java, PHP, and Python applications in a custom SAP billing ecosystem.
- ›Wrote complex PostgreSQL and MySQL queries to debug and fix billing logic, sharpening deep SQL expertise.
Education
Academic background
The degree that started it all.
Projects
What I've built
A mix of open-source personal projects and professional systems built in production environments.
Space Insights
Space data platform on Databricks running two independent fire pipelines: NASA FIRMS thermal detections aggregated into H3 hexagons on a draggable globe, and Sentinel-2 burn-scar detection over four monitored regions, with an unsupervised anomaly model beside the physics-based severity classes. The two measure the same event through different instruments, and a cross-validation model checks one against the other. Fully infrastructure-as-code, zero cloud cost.
OpenClaw
A self-hosted AI assistant I run in production on a GCP VM for my own workflow. The live system is organized around a long-lived gateway that owns sessions, tools, cron jobs, a browser control UI, and Telegram delivery, with all persistent state stored on disk and backed up nightly to GCS. Memory is file-first, with Markdown as the source of truth and SQLite-backed semantic recall layered on top. Inspired the agentic assistant I later built at Atlas Copco.
Lens
Lens is a stock intelligence platform that evaluates companies through five investor frameworks: Graham, Buffett, Lynch, O'Neil and Dividend. A FastAPI backend pulls fundamentals from Alpha Vantage and market data from Polygon, computes transparent weighted scores, and caches analyses in PostgreSQL. The React frontend visualises the result as a radar chart with per-profile metric breakdowns and optional AI commentary. Deployed on GCP with Terraform, Cloud Run, Cloud SQL and GitHub Actions CI/CD.
Memry
Memry is my AI-powered study platform for turning PDFs, EPUBs, and even title-only book lookups into summaries, flashcards, and spaced-repetition review. The backend is a FastAPI service with pluggable LLM, storage, and notification layers; the product ships on both web and mobile. Runs on GCP with Terraform, Cloud Run, Cloud SQL, Cloud Storage, Cloud Scheduler, and GitHub Actions CI/CD via Workload Identity Federation.
IoT Fleet Management Platform
Built an end-to-end fleet platform for industrial compressors spanning embedded software, cloud infrastructure, backend APIs, frontend dashboards, and data engineering. On the device side, I wrote the C++ bridge on Yocto Linux that translates the controller's internal MQTT bus into Azure IoT Hub telemetry, twins, and command flows. In the cloud, I worked on live WebSocket dashboards, remote actions, firmware rollout orchestration, and telemetry storage, with downstream Databricks pipelines feeding fleet analytics and reporting.
Model Predictive Control
Built and productionised a cloud-side MPC stack for industrial compressors. The system combines per-site demand forecasting, mixed-integer optimisation, Databricks-based training and model serving, and a firmware bridge that turns cloud schedules into controller actions. I also worked across the secure device-provisioning path: temporary SAS bootstrap, mTLS certificate issuance, and automated cleanup once devices switch to certificate auth. The result was a control platform that treated latency, deployment safety, and device identity as first-class engineering problems.
Agentic Security Scanner
Azure DevOps security task that detects the project stack, orchestrates multiple scanners, and consolidates findings into one triage flow. An LLM enriches results with exploitability, recommended action, and false-positive filtering. For safe classes of issues it can propose fixes, validate them, and open a PR for human review - built for enterprise CI/CD where remediation speed matters as much as detection.
OpenCloud
Internal team AI assistant for a large industrial company: Slack and Teams chat, Azure DevOps/Databricks/Microsoft 365 integrations via MCP sidecars, persistent team memory, scheduled reporting, and strong operational guardrails around what the agent is allowed to do autonomously.
Finance App
A local-first personal finance tracker built around bank statement ingestion rather than bank APIs. FastAPI handles Excel imports deterministically and uses GPT-assisted extraction for messy PDF statements, then a lightweight category-memory layer learns recurring merchants over time. Everything stays transparent with CSV/JSON storage and a simple review UI.
RAG Knowledge Assistant
proI built and deployed multiple department-specific knowledge assistants for a large industrial company. Each instance used its own Azure AI Search index over internal documents, with scheduled ingestion from systems like SharePoint and Atlassian, Prompt Flow orchestration, and grounded citations in the UI. The interesting engineering work was around multi-tenant isolation, hybrid retrieval quality, evaluation, and operational reliability rather than the chat surface itself.
Databricks Data Platform
proBuilt a production Databricks data platform from scratch for a large industrial company, using a medallion architecture across SQL, JSON, XML, Kafka, and text-based sources. The platform combined PySpark, dbt, Delta Lake, Unity Catalog, and Delta Live Tables to support both batch and checkpointed streaming-style workloads, with governance and schema management treated as first-class design concerns from the start.
Cloud4crc
proA professional FastAPI platform for technician desktop tools, combining file conversion, asynchronous uploads, device registration, and autotune APIs with dual v1/v2 contracts and Azure-native deployment. Much of the design is shaped by backward compatibility for clients that can't always update immediately, rather than by fashion.
Skills
Tech stack
Languages
Data Engineering
AI / ML
Cloud & Infrastructure
Backend
Frontend
Certifications
Verified expertise
Credentials across Databricks, Azure, AWS, and software engineering fundamentals.
Databricks Data Engineering Professional
Databricks
Databricks Data Engineering Associate
Databricks
Azure Data Engineer Associate (DP-203)
Microsoft
Azure AI Engineer Associate (AI-102)
Microsoft
AWS Solutions Architect
Amazon
Professional Scrum Master
Scrum.org
Oracle Java SE 8 Professional
Oracle
Certified Associate in Python
Python Institute
Azure Fundamentals (AZ-900)
Microsoft
Azure Data Fundamentals (DP-900)
Microsoft
Azure AI Fundamentals (AI-900)
Microsoft
Contact
Let's work together
If you're building something ambitious with data or AI, I'd love to hear about it.