Artificial intelligence
AI pipelines · LLMs and RAG · CNN optimisation and computer vision · semantic reasoning and ontologies · behaviour analysis · responsible AI frameworks
Company
About MCi
MCi is a technology company delivering state-of-the-art services in information technology, artificial intelligence, cybersecurity and custom software design. We are a new formation, created from More Group of Companies and Cognity Computing and Consulting SRL — both Romanian companies — and we operate from Bucharest across the European Union and internationally.
Why this formation
The two organisations behind MCi were solving the same problem from different ends.
Cognity Computing and Consulting was built as a research-led engineering practice. Its work concentrated on artificial intelligence and machine learning applied to specialist domains: prompt agents tuned to subject-matter expertise, computer vision for behavioural analysis, differential privacy across the data lifecycle, and cybersecurity compliance treated as an engineering discipline. Cognity's founding principle — compliance by design, ensuring trustworthiness, explainable AI and ethical treatment of bias — carries forward unchanged into MCi.
More Group of Companies brings commercial structure, operational capacity and market reach — the ability to take an engineering capability and deliver it at scale, with the contractual and organisational backbone that serious clients require.
MCi exists because deep technical capability without delivery capacity produces prototypes, and delivery capacity without technical depth produces systems that fail their first audit. We wanted neither.
What we believe
- Systems should be designed for the regime they will live in
- Regulation is not paperwork applied at the end of a project. NIS2, the AI Act, GDPR and ISO 27001 all make architectural demands. Meeting them at design time costs a fraction of retrofitting them later, and produces better systems on every other axis too — better observability, cleaner data boundaries, clearer failure modes.
- AI must be explainable where it matters
- Not every model needs to be interpretable. But a model that influences a medical decision, a legal outcome, a benefit entitlement or a safety system does. We choose architectures accordingly, we audit for bias, and we build human oversight into the workflow rather than claiming it exists in a policy document.
- Privacy is a mathematical property, not a promise
- Removing names from a dataset is not anonymisation — re-identification attacks have proved that repeatedly. We work with formal, provable techniques: differential privacy with calibrated noise, pseudonymisation with proper key separation, and data lifecycle design that treats deletion as a first-class operation.
- Engineering honesty over sales optimism
- We will tell you when a project is scoped wrong, when AI is not the answer, and when the thing you have asked for will not survive contact with your compliance function. This occasionally costs us work. It has never cost us a client.
Leadership
Panos Sampokilis
Founder and Managing Director
Electronics Engineer MSc, IEEE Senior Member, PhD candidate in AI Ethics & Data Privacy (University of Piraeus) and in Explainable AI (University of Athens).
Panos began his career as an electronics engineer in applied research and development, and has spent more than twenty-five years building systems in learning technology, communications, commerce and publishing. As R&D Manager and Technical Director he contributed to the shift in maritime satellite communications toward high-bandwidth systems — architecture deployed across fleets numbering in the thousands of vessels.
He served as Enterprise Architect and AI Lead at the European Commission, DG EMPL, working on EESSI across Member States, coordinating national delegations as single point of contact, and building a responsible AI framework covering bias detection, explainability and ethics review.
His academic research addresses AI, deep learning and the ethical foundations of augmented intelligence. Recent work includes bias auditing in transformer models, differential privacy as a service in AI pipelines, and explainable AI in social security systems.
Capability at a glance
What we work with.
ML / DL stack
TensorFlow · PyTorch · HuggingFace · MLflow · Kubernetes-based MLOps
Engineering
Python · Java · JavaScript · PHP · R · REST API design · microservices architecture · embedded C/AVR
Infrastructure
Linux containers · Kubernetes · Docker · AWS · Azure · enterprise architecture · DevOps and MLOps
Data
PostgreSQL · Neo4j · MySQL · MS SQL · Cassandra · Redis · graph and time-series stores · semantic data stores
Domain
Maritime satellite communications · EU institutional processes · automotive and ECU systems · digital learning platforms · medical signal processing
Where we work
Headquartered in Bucharest, with working presence in Athens and Brussels.
We deliver across the European Union and the EEA, and internationally where the engagement suits remote and hybrid delivery. Bucharest · Athens · Brussels.
Tell us what you're building.
Bring us a defined project, an audit finding, a system that has outgrown its architecture, or a regulation you are not sure how to satisfy. We will tell you plainly whether we are the right people for it.