Machine Learning PhD Intern

Keysight is at the forefront of technology innovation, delivering breakthroughs and trusted insights in electronic design, simulation, prototyping, test, manufacturing, and optimization. Our ~16,800 employees create world-class solutions in communications, 5G, automotive, energy, quantum, aerospace, defense, and semiconductor markets for customers in over 100 countries. Learn more about what we do.

Our award-winning culture embraces a bold vision of where technology can take us and a passion for tackling challenging problems with industry-first solutions. We believe that when people feel a sense of belonging, they can be more creative, innovative, and thrive at all points in their careers.

Keysight's AI team develops machine learning and optimization methods for engineering problems where simulation, measurement, and intelligent design workflows interact.

This internship focuses on rigorous benchmarking of machine learning and optimization methods for inverse design and optimization. Meaningful comparisons require more than running different algorithms on the same problem: objectives, constraints, initialization, evaluation budgets, simulation settings, and statistical procedures must be controlled carefully.

You will develop a reproducible benchmarking framework for comparing an existing engineering design workflow with selected academic and open-source methods across representative inverse design and optimization problems.

As a PhD Intern in Machine Learning and Optimization, you will design, implement, and evaluate a benchmark suite covering a small number of representative engineering design problems.

The work spans the full benchmarking process: reviewing relevant research, selecting strong comparison methods, defining fair experimental protocols, integrating available implementations, automating simulation-based experiments, and analyzing where different approaches succeed or fail.

A central objective will be to distinguish meaningful algorithmic differences from effects caused by evaluation budget, initialization, training cost, randomness, or implementation choices. You will measure both the quality of the final engineering solution and the computational or simulation effort required to obtain it.

The internship will focus on two agreed engineering use cases and approximately two or three carefully selected baseline methods. The goal is depth and experimental rigor rather than a broad but shallow algorithm survey.

By the end of the internship, you will deliver a reproducible benchmarking workflow, quantitative comparison results, and an evidence-based technical report describing performance, robustness, limitations, and areas requiring further evaluation.

You will work alongside machine learning researchers, optimization experts, and simulation engineers.

• Review relevant state-of-the-art research and available open-source implementations in inverse design, black-box optimization, reinforcement learning, and related methods as appropriate to the selected use cases.
• Select a small set of technically meaningful baseline methods together with the supervising researchers.
• Define representative engineering optimization problems with explicit design variables, objectives, constraints, parameter ranges, and simulation settings.
• Establish matched experimental protocols covering initialization, random seeds, evaluation budgets, stopping criteria, constraint handling, and training or model-building costs.
• Adapt and integrate baseline implementations with existing engineering simulation workflows.
• Automate repeated experiments using reproducible scripts and configuration-driven workflows.
• Measure final design quality under fixed engineering evaluation budgets.
• Quantify sample efficiency, including the number of simulation or engineering evaluations required to reach defined performance targets.
• Compare target satisfaction, objective quality, constraint violations, wall-clock cost, and variability across repeated runs.
• Account explicitly for initialization, surrogate-model fitting, policy training, or other algorithm-specific costs where relevant.
• Evaluate robustness across random seeds, problem variations, noisy observations, and constrained design spaces.
• Investigate where methods perform consistently, where results become unstable, and where conclusions depend strongly on the selected problem or budget.
• Evaluate reuse or transfer of information across related design problems where appropriate.
• Record configurations, software versions, random seeds, datasets, and experimental results so that comparisons can be reproduced.
• Produce reusable benchmark scripts, analysis tools, plots, and tables.
• Document software sources, dependencies, and licensing information for internal review.
• Present findings with clear separation between experimentally supported conclusions, observed limitations, and areas requiring further study.

• Apply modern optimization and machine learning methods to realistic engineering design problems.
• Learn how to design statistically sound benchmarks for simulation-based optimization and inverse design.
• Develop practical understanding of sample efficiency, robustness, convergence behavior, and solution quality.
• Learn how evaluation budgets and experimental protocol choices can materially affect conclusions about algorithm performance.
• Gain experience assessing research methods for reproducibility, applicability, and engineering relevance.
• Build working fluency with simulation-driven optimization and engineering design workflows.
• Develop experience automating large experimental studies and communicating uncertainty and limitations clearly.
• Take a benchmark methodology from research experiments to a reusable engineering evaluation framework.

• Currently pursuing or recently completed a PhD in Machine Learning, Computer Science, Electrical or Electronic Engineering, Applied Mathematics, or a related technical field.
• Strong Python programming skills, including scientific computing, data analysis, and experiment automation.
• Strong understanding of numerical optimization, machine learning, experimental design, and statistical evaluation.
• Experience implementing, adapting, or evaluating optimization or machine learning algorithms beyond tutorial-scale examples.
• Familiarity with at least one relevant class of methods such as Bayesian optimization, evolutionary or global optimization, surrogate-assisted optimization, policy optimization, or model-based reinforcement learning.
• Ability to read research papers critically, understand methodological assumptions, and translate published methods into reproducible experiments.
• Experience with repeated experimental evaluation, multiple random seeds, controlled comparisons, and quantitative analysis.
• Working knowledge of probability, statistics, and the interpretation of variability across experimental runs.
• Experimental discipline, including careful baselines, reproducibility, experiment tracking, and transparent reporting of inconclusive or negative results.
• Willingness to understand the underlying engineering problem rather than treat it as an abstract optimization benchmark. Mentorship in the domain will be provided.
• Clear technical communication and the ability to collaborate with researchers and engineers from different technical backgrounds.

• Familiarity with Git, modular software development, and reproducible experiment workflows.
• Comfort working with existing research code and simulation interfaces rather than developing exclusively from clean-room examples.
• Understanding of experiment tracking, statistical comparison, and numerical reproducibility.
• Ability to design an evaluation that answers a technical question rather than simply produces a favorable number.
• Ability to communicate assumptions, limitations, and uncertainty clearly.