Algorithmic Chain Development Engineer

The University of Technology of Compiègne (UTC) is seeking an algorithm chain development engineer to join its Applied Mathematics Laboratory (LMAC) - Computer Engineering Department (GI), as part of the ANR INCA project.

Mission

Contribute to the development of a complete algorithmic chain for CSM/CSMC (from modelling to reproducible code) in order to handle real cases and strengthen internal/external collaborations.

Activities

• Modelling and formalisation: specification of spatio-temporal CSM/CSMC (explicit durations, space-dependent transitions, covariates, decision policies).

• Parametric and non-parametric inference: development of maximum likelihood estimation algorithms (generalised EM/ECM), decoding (semi-Markovian Viterbi), filtering/smoothing (forward–backward with durations), model selection (informational criteria, cross-validation).

• Optimisation and control: formulation and resolution of decision support problems (semi-Markovian MDP/POMDP), policy calculation and sensitivity analysis.

• Simulation and calibration: spatio-temporal simulators, numerical experiments, robustness study (noisy/missing data), uncertainties (bootstrap, likelihood profiles).

Scientific context and objectives

Semi-Markov chains (SMCs) and hidden semi-Markov chains (HSMCs) have long been a central focus of stochastic modelling within our laboratory. Our work has produced theoretical and methodological advances (asymptotic, statistical estimation), but relatively few algorithmic and software results — a shortcoming that limits the value of our contributions and synergies with other teams. As part of the ANR-INCA project, we are recruiting an engineer to design, analyse, implement and validate processing and estimation algorithms for CSM/CSMC, with a priority application to the spatio-temporal modelling of epidemic propagation (e.g. COVID-19). The issues include in particular:

Regional time of arrival of an epidemic outbreak initiated in a large urban area, depending on geography and mobility;

Decision support under uncertainty (which interventions to take from a given state to minimise risk).

Additional information

Type of contract and expected start date

Fixed-term contract - expected duration of 12 months - to be filled as soon as possible and until 31/12/2026 at the latest

Gross monthly salary

Depending on funding

Hours

37 hours and 30 minutes/week - 1,607 hours/year

Working environment and context

The successful candidate will work within the LMAC/GI Applied Mathematics Laboratory and will strengthen the team in its research and development activities in a stimulating and interdisciplinary scientific environment. The research team, which is involved in several national projects, offers a framework conducive to scientific advancement.

This recruitment is funded by the French National Research Agency (ANR).

Skills

• Excellent command of Markov and semi-Markov processes/chains, including hidden versions (HSMM) and associated inference methods.

• Solid foundation in statistics (likelihood, EM, Fisher information), probability, numerical optimisation, computational linear algebra.

• Proven experience in scientific programming: Python (NumPy/SciPy, JAX or PyTorch as needed), R (data.table, Rcpp), and ideally C++ (profiling/optimisation).

Qualifications, field of study

• Qualifications: Master's degree in research/engineering (M2); PhD preferred.

• Field of study: applied mathematics, statistics, probability/Markov chains, statistics/inference, optimisation, scientific programming.