Skip to main content

f-inference

ABOUT / OBJECTIVES

Five objectives. One ambitious shift.

Connecting design-time optimisation, adaptive execution, trustworthy orchestration and demonstrable real-world impact.

RESEARCH OBJECTIVES

What we are working towards.

OBJECTIVE 01

Optimise foundation models at design time

Develop strategies for profiling, early exits, compression and distributed model adaptation.

OBJECTIVE 02

Execute inference efficiently

Use dynamic routing, distributed models, adaptive caching and lower-overhead communication.

OBJECTIVE 03

Guide the move beyond the cloud

Balance techno-economic constraints, security, performance and data sovereignty.

OBJECTIVE 04

Validate through industrial experiments

Evaluate technologies in digital industry, automotive and healthcare.

OBJECTIVE 05

Maximise research and societal impact

Publish open tools, engage with standards, and connect European and international communities.

Measuring progress

Research and pilot evaluation will consider latency, energy consumption, reliability, security, affordability and policy constraints.

Quantified targets in the source document require coordinator approval before public publication and are deliberately omitted from this demo.