References

Real-world projects in industry, research, and finance.

ArcelorMittal

Industrial process control

Context

A tube and pipeline manufacturing plant for the oil industry needed a comprehensive process control system covering the entire production chain.

Engagement

  • Raw materials logistics management on the storage yard
  • Production tracking for each station on the line
  • Complete product traceability
  • Integration of multiple industrial buses
  • Interfaces on industrial panels, PCs, and Android tablets
  • Server and network infrastructure (Ethernet, fiber optics, ring topology)
  • Custom driver development

Result

Operational solution successfully deployed. The group subsequently ordered its transposition to other plants in Eastern Europe and the Middle East.

2+ years C C++ Pascal Erlang/OTP WinDev CoDeSys Assembler SQL Server SQL

ArcelorMittal

Vision-based predictive maintenance

Context

Certain wear parts required regular replacements. The challenge: anticipating these replacements to avoid unplanned production stops.

Engagement

  • Camera installation filming wear parts
  • Machine learning algorithms analyzing the video feed
  • Automatic wear level detection
  • Real-time alerts for maintenance teams

Result

Deployed solution enabling maintenance needs anticipation and intervention optimization.

ML / Vision Python C# .NET

Airbus Helicopters

Reverse engineering & critical integration

Context

A Eurocopter subcontractor needed to integrate two Italian vertical storage machines (Modula) into a helicopter blade production line. A commercial dispute prevented obtaining technical documentation. Delay penalties were accumulating.

Engagement

  • Complete reverse engineering of the Modula communication interface, without documentation
  • Communication gateway between Schneider PLCs and Italian electronics
  • Choice of Erlang to guarantee high availability
  • Testing and validation in a secure environment (military production)

Result

Operational gateway, project unblocked. The solution was subsequently maintained for several years.

High availability Erlang Aerospace

Top engineering school

AI-powered bibliographic research platform

Context

Faculty researchers were spending considerable time on bibliographic research: identifying relevant sources, reading and synthesizing articles, cross-referencing results across disciplines. With the emergence of LLMs, significant automation potential was identified — provided the methodological rigor of scientific research was respected.

Engagement

  • In-depth business interviews with researchers to understand their decision-making processes
  • Identification of automatable steps and formalization of a generic methodology
  • Agentic architecture: specialized agents (search, analysis, synthesis) coordinated by an orchestrator
  • RAG on scientific corpus: article vectorization, semantic search, sourced synthesis generation
  • Heterogeneous data ingestion pipelines (PDF, bibliographic databases, academic APIs)
  • Proactive contribution on user interfaces and researcher experience

Status

Project currently in development.

Ongoing Research LLM RAG Embeddings Multi-agents Go TypeScript Bun SQLite

Trador

Algorithmic trading platform

Context

An entrepreneur marketing an automated trading solution needed to urgently migrate his platform to MetaTrader 5, as his provider was making the current solution obsolete.

Engagement

  • Rapid understanding of trading business logic
  • Architecture design for the new solution
  • MQL5 trading logic development
  • Backend connected to MetaTrader 5
  • Cross-platform dashboard (web, mobile, desktop)

Outlook

Enrichment of the solution with machine learning and artificial intelligence algorithms.

Trading MQL5 C++ ZeroMQ Go TypeScript React Native PostgreSQL Multi-tier architecture

Luma

Technical partnership

Context

A personal "life copilot" project, combining data aggregation (health, wearables, psychological profile, context) and conversational AI for personalized advice.

Our role

THE LOOP is the technical partner and future associate in the Belgian company that will operate the project. This model illustrates our ability to commit beyond simple service delivery.

Partnership AI LLM Embeddings Agentic architecture Go Dart Flutter PostgreSQL Full-stack

Other projects

Cheating detection — CNRS

Implementation of a cheating detection solution in online card games, based on the laboratory's research work.

Sentiment analysis — Streaming

Analysis of viewer emotional intensity via online stream chat, to optimize real-time ad placement.

Pollutant classification

Mentoring a PhD student on integrating ML algorithms for classifying indoor air pollutant gas families.

IoT for research

Multiple IoT developments for a top engineering school: backends, server infrastructure, embedded development (Arduino, ESP32).