> Markdown version of [/jobs/ext/4600754-lead-rf-data-scientist](https://www.wearedevelopers.com/jobs/ext/4600754-lead-rf-data-scientist). 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).

---

# Lead RF Data Scientist

- **Company:** Forterra
- **Location:** Durham, NC, United States
- **Experience:** Expert
- **Salary:** $175,000 - $220,000
- **Contract:** Permanent contract
- **Skills:** Artificial Neural Networks, C++ (Programming Language), Nvidia CUDA, Data Transformation, Formal Verification, Python (Programming Language), NumPy, SciPy, Signal Processing, Management of Software Versions, Pytorch, Evaluation Pipelines, Knowledge Representation, Information Technology, Data Generation
- **Published:** November 6, 2026
- **Apply:** https://www.jofdav.com/jobs/60438995-lead-rf-data-scientist

## About the Role

M.S. or Ph.D. in electrical engineering, statistics, applied mathematics, computer science, physics, or a related quantitative discipline, or equivalent experience.3+ years of applied research or product development using measured data, including ownership of experiments from hypothesis through evaluation and release recommendation.Strong foundations in statistical learning, estimation, optimization, linear algebra, numerical methods, and experimental design. Able to reason about uncertainty, identifiability, calibration, and failure modes rather than only optimize a leaderboard metric.Demonstrated experience building reproducible training and evaluation pipelines, including dataset versioning, data-quality controls, held-out evaluation, and experiment tracking.Proficiency in Python and the scientific Python ecosystem (for example NumPy, SciPy, PyTorch or JAX). Able to read C++ and CUDA sufficiently to verify that the deployed numerical behavior matches the intended
technique contract.Able to author a technique specification, an evaluation plan, and a clear evidence-backed recommendation that another technical reviewer can audit.Strong written and verbal communication; able to work directly with engineers, analysts, and operational users without diluting technical rigor.Must be a U.S. person eligible to obtain and maintain a U.S. security clearance.

Preferred Qualifications
RF signal-processing experience with complex IQ, spectrograms, modulation, radar waveform parameters, receiver/channel impairments, and multi-signal interference; or a strong signal-processing background with the ability to learn RF domain specifics quickly.Experience with source separation, multi-target tracking, structured prediction, time-frequency analysis, inverse problems, Bayesian methods, or uncertainty-aware deep learning.Applied research experience on national-security, intelligence-analysis, autonomy, communications, sensing, or other high-consequence operational problems.Experience building GPU-backed techniques or numerical pipelines in PyTorch, JAX, CUDA, or equivalent tooling.Human-in-the-loop labeling, active learning, data-centric methods, or evaluation of partially labeled real-world corpora.Ontology engineering, knowledge representation, or mapping technique outputs to structured semantic or event systems.Experience with technique release governance,
model cards, safety cases, reproducible science, or formal verification of numerical software.

## Description

We are hiring a technical research lead who wants to make a visible impact. You will take ambitious ideas from mathematical formulation through working techniques, testing, and operationally relevant evaluation. You will work closely with RF and DSP engineers, platform engineers, and analysts who use the results. The goal is not research for its own sake: it is to create capabilities that are useful, understandable, and dependable when they matter.

This is a hands-on, mid-career applied-research role for someone who enjoys moving between signal processing, statistical learning, numerical methods, and experimental design. You should be excited by hard, imperfect data; eager to learn from failures; and motivated to see your work shape real systems.

What you'll do
Lead RF technique development for detection, classification, parameter estimation, and understanding complex signal environments. Choose methods that fit the problem rather than treating a neural network as the default answer.Define technique inputs, targets, loss functions, uncertainty outputs, and decision rules for real RF collections. Account for sample rate, bandwidth, channel and receiver effects, incomplete labels, interference, and overlapping or confusable signals.Build a disciplined data practice: versioned datasets and splits, label quality rules, leakage checks, and reproducible training and evaluation recipes. Make synthetic, simulated, and collected data distinguishable in every result.Design synthetic and hybrid data generation that exercises the technique's intended operating envelope, then characterize where synthetic performance does and does not transfer to collected data.Run training and ablation studies; diagnose failures as data, labeling, representation,
optimization, technique, or evaluation problems; and make the next experiment smaller and more informative.Establish meaningful technique qualification: held-out and stress evaluation, calibration, confusion analysis, uncertainty behavior, tail failures, and clear criteria for whether a technique is ready to move forward.Own the mapping from learned signal behavior to consistent, structured event and object descriptions, with explicit evidence rules, so that technique outputs can be understood and acted upon.Work with runtime engineers to specify technique artifacts, features, preprocessing, numerical tolerances, test vectors, and fallback behavior. Review C++/CUDA implementation changes for fidelity to the technique specification.Work directly with analysts and field teams to turn collection observations into prioritized labeling, replay, and evaluation work. Use those findings to improve the next technique or to narrow its approved operating envelope.Produce durable technical
artifacts: derivations, experiment plans, technique documentation, data and label specifications, evaluation reports, and concise design reviews that distinguish measured evidence from assumptions.Travel up to 10% to collection sites and test events, with the ability to commute to Durham (RTP), NC or Arlington, VA as needed.