> Markdown version of [/jobs/ext/4176258-user-terminal-principle-modem-waveform-architect-oneweb-next](https://www.wearedevelopers.com/jobs/ext/4176258-user-terminal-principle-modem-waveform-architect-oneweb-next). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# User Terminal Principle Modem Waveform Architect - OneWeb Next

- **Company:** Eutelsat
- **Location:** London, UK
- **Contract:** Permanent contract
- **Skills:** 4G (Telecommunication), Abstraction Layers, C++ (Programming Language), Code Review, Protocol Stack, Modems, Data Transmissions, Ethernet, Firmware, Field-Programmable Gate Array (FPGA), Integrated Development Environments, Python (Programming Language), MATLAB, Machine Learning, Multiprocessing, Reduced Instruction Set Computing, Signal Processing, Subsystems, Toolchain, GNSS
- **Published:** October 27, 2026
- **Apply:** https://www.indeed.com/viewjob?jk=9614329b1d2a8790

## About the Role

* MSc or PhD in signal processing, digital communications, telecommunications or electronics;
 * 12+ years in digital modem or wireless physical layer engineering, including responsibility for the algorithmic architecture or the technical acceptance of a product taken through silicon or FPGA to measured field performance;
 * Deep expertise in synchronisation and estimation theory applied to real receivers: acquisition at low SNR, CFO and SFO estimation, timing recovery, phase tracking under phase noise, channel estimation and equalisation;
 * Practical command of forward error correction: LDPC, turbo or polar decoding architectures, iteration and quantisation trade-offs, and their cost in cycles, memory and power;
 * Hands-on experience of programmable PHY platforms and their SDKs: micro-kernel or kernel-library based signal processing, hardware abstraction layers over accelerators, and the bring-up of a PHY on a vendor development environment;
 * Proven experience of fixed point design and of quantifying implementation loss, quantisation noise and dynamic range against a floating point reference;
 * Working knowledge of real-time partitioning across heterogeneous cores and accelerators, and of the cycles-per-sample, memory bandwidth, DMA and latency budgets that constrain it;
 * Ability to build and own a bit accurate simulation chain in Matlab or Python, and strong C/C++ with vector intrinsics on embedded targets;
 * Command of RF impairment modelling and its budgeting: phase noise, IQ imbalance, DC offset, PA non-linearity, converter noise and spurious;
 * Laboratory characterisation with vector signal generators and analysers, channel emulators and timing test sets, and the ability to reconcile measurement with simulation;
 * Proven ability to form an independent technical judgement on a design you did not author, to challenge it constructively, and to substantiate your position with analysis, simulation or measurement rather than opinion;
 * Intellectual rigour and curiosity: the instinct to ask the question that exposes a weak assumption, and the willingness to be proven wrong by the data;
 * Comfortable building a capability rather than inheriting one: little existing tooling or process to rely on, decisions to be taken with incomplete information, and the autonomy and resilience that this demands;
 * Eligibility under applicable European and United States export control regulations;
 * Fluent professional English.

Assets

 * PhD with published work in synchronisation, estimation, detection or coding theory;
 * Experience with LEO or multi-orbit satellite terminals, gateways or payload processing;
 * Knowledge of 3GPP non-terrestrial network specifications from Release 17 onwards, or of DVB-S2X and DVB-RCS2;
 * Experience with multi-core RISC-V subsystems, vector DSP engines and micro-kernel firmware, including toolchains, interrupt controllers, DMA engines and inter-processor communication;
 * Exposure to converged 4G, 5G and 6G modem platforms supporting carrier aggregation, FR1 and FR2, and MIMO configurations up to 8x8;
 * Experience of technical due diligence on silicon or IP suppliers, or of acceptance of externally developed physical layer software;
 * Interest in machine learning applied to the physical layer: neural receivers, learned channel estimation or interference mitigation, and their deployment on embedded NPUs;
 * Knowledge of electronically steered antennas: element calibration, beam pointing, and the interaction between array behaviour and the physical layer;
 * Converter and clocking design, PTP and the G.8275 profiles, G.8273.2 clock classes, APTS and GNSS assisted timing, and holdover behaviour;
 * Experience of mentoring engineers and of setting technical practice in a team being built from scratch.

## Description

* Master the platform hands-on: SDK and toolchain, micro-kernel architecture of the PHY, mapping onto signal units, scheduling and synchronisation across units, hardware abstraction layers over the FEC and signal processing engines, DMA fabric, interrupts and inter-processor communication. You should be able to build, modify, instrument and profile PHY code yourself;
 * Own the terminal physical layer requirements and their traceability, from service and constellation constraints down to waveform, timing and modem parameters, and hold the specification stable through the programme;
 * Assess and challenge the receive chain design: acquisition strategy and its performance below 0 dB Es/N0, CFO and SFO estimation, timing recovery, carrier phase tracking under composite LO and converter phase noise, channel estimation and interpolation, equalisation, soft demapping and LLR scaling, AGC dynamics, and the loop bandwidth and stability margins behind each of them;
 * Verify the coding scheme and its operating point: code family and rate set, decoder mapping onto the FEC accelerator, iteration budget and early termination, LLR quantisation, and the actual position of the BLER waterfall and error floor relative to the capacity bound;
 * Scrutinise the transmit chain: pulse shaping and spectral mask compliance, PAPR and crest factor reduction, digital pre-distortion interface and training path, and the consistency of the EVM, ACLR, spurious and EIRP budgets from baseband to antenna aperture;
 * Interrogate the Doppler, delay and mobility strategy: ephemeris and GNSS aided pre-compensation of frequency and timing, closed loop refinement and timing advance, sampling frequency offset correction, the split of the residual error budget across GNSS, oscillator and tracking loops, and terminal behaviour through make-before-break transitions, re-acquisition and outage recovery;
 * Review the timing and clocking architecture and hold it to its targets: oscillator class and holdover drift, phase noise and jitter budget across reference, PLL, converter clocking and SYSREF distribution, GNSS and 1PPS handling, PTP client, boundary clock and grandmaster behaviour under the G.8275 profiles, and the max|TE|, MTIE and TDEV limits of the applicable G.8273.2 clock class;
 * Audit the fixed point design: wordlength and scaling choices, saturation and dynamic range handling, and the per-block implementation loss budget against a floating point reference;
 * Scrutinise the real-time partitioning across vector compute engines, accelerators and RISC-V control cores: cycles per sample and per slot, scratchpad allocation, DMA and double buffering, memory bandwidth against LPDDR or DDR, the deterministic latency chain, and the margin left for growth;
 * Question the co-residency of the protocol stack and host processor functions on the spare cores of the same SoC: real-time isolation, contention on shared memory and interconnect, and the margin policy that follows;
 * Assess where on-chip machine learning inference genuinely improves receiver performance, and under what complexity, latency and validation conditions such a function would be acceptable in an operational terminal;
 * Review the modem interfaces: JESD204D/E lane rates, subclass and deterministic latency, analogue IQ alternatives, sampling and clocking plan for GHz-class integrated converters, antenna control and calibration signalling, Ethernet and fronthaul, and the boundary towards the protocol stack and host processor;
 * Build and own an independent bit accurate reference model and the golden vectors derived from it, and use them as the basis for acceptance rather than relying on figures supplied to us;
 * Define the acceptance criteria and the milestone entry and exit conditions, then run or supervise the acceptance campaigns: BER and BLER against Es/N0, acquisition time distributions, sensitivity to Doppler rate and phase noise, handover interruption time, EVM and spectral compliance, timing against the applicable masks, and model to measurement correlation in the laboratory and over the air;
 * Lead joint architecture and code reviews with partner engineering teams, arbitrate on technical merit, and secure that design rationale, models, documentation and know-how land inside Eutelsat in a form our own engineers can use;
 * Protect the portability of the physical layer software across silicon generations, so that the investment made in micro-kernels, models and test assets survives the move to the next platform;
 * Build the in-house competence: define the modelling, tooling and review practices the team will work by, help recruit and technically lead the engineers who join, and grow into the architecture lead of that group.