Embedded AI / 5G Engineer, O-RAN - Paris - H/F

Free Mobile, part of the Iliad group, designs and deploys its own 5G infrastructure, from O-RAN radio units and distributed units to the user plane and Cloud RAN environment.

At the heart of the DU, the physical layer and 5G fronthaul operate under hard real-time constraints. Every 0.5 ms slot must be received from the radio unit over eCPRI, processed on GPU, DSP or SIMD CPU, and delivered to the upper layers on time.

This role sits at the intersection of 5G signal processing, embedded software and high-performance networking. You will work with real IQ data, radio units and production constraints, not only simulations.

AI-assisted algorithms are part of the toolbox, but not the objective by themselves. We use them when they deliver a measurable improvement over classical methods in BLER, throughput or latency.

Your role :

You will develop and optimise physical-layer and fronthaul processing for an O-RAN 7.2x system, from algorithm design and simulation to real-time implementation and integration with radio units.

Your responsibilities will include:

• Developing the 5G uplink receiver chain: channel estimation, interpolation, MMSE equalisation, MIMO detection, LLR computation and LDPC decoding.
• Implementing PRACH detection, timing estimation and beamforming for massive MIMO.
• Processing O-RAN 7.2x C-plane and U-plane traffic, including eCPRI timing windows and IQ compression.
• Implementing real-time L1 functions in C on Linux and accelerating compute-intensive kernels with CUDA, DSP or SIMD.
• Building the eCPRI datapath with DPDK using poll-mode I/O, rte_flow, dedicated queues, zero-copy buffers and hardware timestamps.
• Optimising the complete processing chain for latency, memory usage, cache behaviour and compute load within a 0.5 ms slot budget.
• Validating algorithms through link-level simulations using 3GPP TDL/CDL channel models, then testing them with real IQ captures and O-RAN radio units.
• Measuring BLER, throughput, latency and CPU, GPU or DSP load to compare classical algorithms with AI-assisted alternatives.
• Investigating topics such as neural channel estimation and neural receivers when they can outperform an established baseline.
• Building regression tests and CI to detect performance or accuracy degradation.
• Contributing to architecture decisions, including GPU versus DSP or SIMD, hardware selection and accelerator integration.
• Working closely with radio unit, RAN, core network and operations teams.

You will progressively own complete technical topics, from the mathematical model and implementation choices to their validation on the DU test bench.

You will also contribute to code reviews, design documentation and open-source projects such as DPDK, Linux and U-Boot.

Your profile :

We are open to both experienced engineers and high-potential graduates with strong foundations in mathematics, signal processing or embedded software.

You have:

• An engineering degree, Master of Science or PhD in telecommunications, applied mathematics, embedded systems, computer science or a related field.
• Strong foundations in linear algebra, probability and digital signal processing.
• A good understanding of OFDM, MIMO, Fourier transforms, channel estimation, detection or channel coding.
• C programming skills on Linux, with an interest in memory layout, multithreading, cache behaviour and performance.
• Python experience with NumPy and SciPy for simulation, testing and data analysis.
• An interest in at least one of the following areas: 5G/LTE physical layer, O-RAN fronthaul, real-time embedded software, GPU programming, DSP/SIMD optimisation or AI for signal processing.
• The ability to turn mathematical concepts into code, define relevant metrics and validate the results experimentally.
• Good written English for reading specifications, documenting designs and reviewing code. French is a plus.

Experience with CUDA, cuBLAS, cuFFT, AVX2, AVX-512, Arm NEON/SVE or fixed-point arithmetic is valuable, but you do not need to master all of them.

Knowledge of DPDK, eCPRI, PTP, O-RAN 7.2x, 3GPP channel models or massive MIMO is also a plus.

What matters most is your ability to understand difficult technical problems, build measurable solutions and explain the choices behind them.

Recruitment process :

• 15-minute phone screen with Steve, the recruiter
• 30-minute interview with Vincent, the hiring manager
• 60-minute technical interview
• 60-minute HR interview with Sarah and Steve

Benefits :

• Profit-sharing and incentive scheme based on the Group’s performance
• Reimbursement of your Free home internet plan
• One to two remote-working days per week
• Secure bicycle parking
• On-site restaurants and cafeteria
• Numerous internal mobility opportunities across the Iliad Group
• Regular training opportunities and access to Free University to develop your skills
• Unique workspaces located in an iconic building that is part of Paris’s architectural heritage
• Public transport costs reimbursed up to 70%
• Swile meal card worth €10 per working day, with 60% covered by the company
• Comprehensive health and income protection coverage
• Time Savings Account, known in France as a Compte Épargne Temps or CET
• Employee benefits through the CSE, including gift vouchers, discounts and exclusive offers

Le groupe Iliad s’engage pleinement en faveur de ses collaborateur·rice·s, à travers une politique stricte de non-discrimination et de lutte contre le harcèlement. Toutes les candidatures sont les bienvenues, quels que soient le parcours, l’âge, l’origine, le genre ou l’orientation sexuelle.