Sensor Data Analysis and Model Fit (MATLAB)

We study how professionals actually work - how experts drive creative and technical software - as research toward better computer interfaces. This is a paid demo project: you complete one realistic task while we record the screen and your spoken commentary. This is the first round. If it goes well, there is a real chance it turns into ongoing work. Assignment Take a noisy sensor time series, characterize the noise, clean it, fit a model that explains the underlying signal, validate that model honestly, and write it up as a short technical report with figures. You source or synthesize the data yourself. Time: 3–4 hours, in one non-stop sitting Timing: we need the session completed this week You bring your own data Source a real noisy sensor dataset or synthesize a realistic one, and say where it came from. How the session works You work inside a cloud computer we provide, with the hardware and software already set up. Your screen, mouse, keyboard and microphone are recorded for the whole session, and you talk through what you are doing as you do it. We are recording how you work, not grading how closely you follow instructions. Use your own workflow and research the way you normally would. Dead ends, restarts and changed minds are genuinely useful to us - please do not tidy up your process. Spoken English matters for this one The commentary is half of what we are paying for, so you need to be comfortable thinking out loud in English for a few hours straight. Fluency matters far less than being able to keep talking while you work. So please include a short voice recording with your proposal - under 60 seconds, but long enough for us to hear how you explain something in English - answering this question: "When you open a noisy signal, what do you look at before you filter anything?" Just your honest answer in your own words. We are listening for two things: how you explain your workflow, and whether we can follow your spoken English. It does not need to be a perfect or complete answer. Proposals without a voice note will not be reviewed. Deliverables - Runnable scripts (.m or .py) that reproduce everything end to end. - The dataset you used, or the script that generates it. - Exported figures, each with labelled axes and units. - A short technical report, MAT-001_report.pdf, covering method, results and limitations. - A Working_Notes folder with your references, sketches, scratch files and notes Ideal freelancer Engineer or scientist comfortable with signal processing, model fitting and honest validation.