Expertise
Eight areas, one engineering standard
We work where technology has to produce a measurable effect. Each area below is a capability we run end to end — from framing to production and handover.
- 01
Digital public service platforms
Turning administrative procedures into working online services, and making the relationship between institutions and citizens flow.
We build platforms that carry a procedure from end to end: submission, identity verification, processing by case officers, hierarchical approval, payment and issuance of the document. The architecture assumes irregular connectivity and teams for whom the digital tool is new.
What this covers
- Paperless procedures and supporting documents
- Approval workflows and electronic signature
- Mobile money and payment integration
- Appointment booking and queue management
- Steering dashboards and activity statistics
- Citizen portals usable on an entry-level phone
- 02
Remote sensing, satellite imagery and AI-driven geo-intelligence
Processing satellite imagery and field data, combining them with machine learning, and producing geographic information a decision-maker can act on.
We work with Earth observation programmes — optical and radar imagery, time series, open data — and combine them with machine learning models to produce a continuous reading of a territory: land cover, crop condition, water stress, urban dynamics, coastal erosion, risk exposure. Where manual photo-interpretation would take weeks, the processing chain returns a dated, comparable, mapped result. Deliverables are interactive maps, alerts and periodic reports designed to be read by non-specialists.
What this covers
- Optical and radar remote sensing, multi-date time series
- Pre-processing, atmospheric correction and spectral indices
- Classification and change detection through machine learning
- Interactive mapping and geographic information systems
- Crop monitoring, water stress and food security
- Urban growth, coastline, natural resources and risk exposure
- Early warning systems and periodic territorial reports
- 03
Artificial intelligence and computer vision
Building models that are useful, measurable and deployable under real operating conditions.
Our work runs from qualifying the need to production: assembling and annotating datasets, selecting and training models, evaluating them against business criteria, then industrialising. We favour models that run on modest hardware and, where possible, directly on the device, without a permanent network dependency.
What this covers
- Computer vision: detection, recognition, image analysis
- Natural language processing and voice interfaces
- Assistive technology and digital accessibility
- Predictive models and decision support
- Industrialisation, monitoring and model retraining
- Bias evaluation and a responsible-use framework
- 04
Digital health and e-health
Equipping care facilities and public health programmes with reliable applications that work in poorly connected areas and respect medical confidentiality.
Health data sets a particular bar: it must be accurate, traceable, available at the right moment and protected at all times. We build patient monitoring, facility management, prevention campaign and epidemiological collection applications, designed to work offline and synchronise as soon as the network returns. Every project is run with the health professionals concerned, because a badly designed record is a record nobody fills in.
What this covers
- Patient records and longitudinal care pathways
- Offline field applications for community health workers
- Epidemiological collection, surveillance and public health dashboards
- Appointments, reminders and prevention campaigns
- Clinical decision support and medical image analysis
- Confidentiality, encryption, access logging and controlled hosting
- 05
Environmental monitoring through artificial intelligence
Having computer vision automatically recognise what must be counted, sorted, monitored or reported in the environment.
Counting by hand does not scale, and estimating without counting does not convince. We build recognition chains that produce a dated, comparable figure — and that say what they cannot see, because a detection model with an unstated error rate is an opinion, not a measurement.
What this covers
- Object detection and classification applied to waste sorting
- Illegal dumping detection and urban cleanliness monitoring
- Automated counting and material flow quantification
- Natural environment and biodiversity monitoring
- Air and water quality, connected sensor networks
- Environmental dashboards and compliance reporting
- 06
Traceability, quality and operations management
Making what actually happens in the field visible and verifiable: what was done, by whom, when, where and to which standard.
Proof of execution is worth more than a declaration of execution. We build chains that timestamp, locate and document each step, so that a dispute is settled by a record rather than by a recollection.
What this covers
- Product and supply chain traceability, from origin to delivery
- Quality control, inspection plans and non-conformity records
- Field intervention tracking and timestamped proof of execution
- Asset, stock and maintenance management
- Performance indicators and team management
- Tamper-evident logging and audit trails
- 07
Mobility and territorial safety
Equipping authorities to understand, prevent and handle events in the field.
These systems are judged on the worst day, not the average one. We design them so that a field agent can work without a network, so that the control room sees the situation rather than a table, and so that the official document produced at the end holds up.
What this covers
- Field mobile applications, including offline
- Real-time supervision and heat mapping
- Production and tracking of official documents
- Geolocated notifications and alerts
- Interconnection with existing information systems
- Statistical analysis and activity reporting
- 08
Advisory, engineering and training
Supporting organisations in their technology choices and making their teams self-sufficient.
The most useful engagement is sometimes the one that stops a project. We audit, we frame, we cost — and we say what does not fall to us before the signature, not during the work.
What this covers
- Digital maturity audit and strategic roadmap
- Project framing, estimation and specification writing
- Software architecture and code review
- Cybersecurity and data protection support
- Training in artificial intelligence, data and development
- Skills transfer and team support
A need that cuts across several of these?
That is the most common case. We scope the whole before splitting the work — and we tell you what does not fall to us.