Welcome to Marcel Merchat's Website
RF Electronic Engineer and Data Scientist
Chicago Metropolitan Area, Illinois
I seek work as a data scientist and RF communication engineer. I have worked in the two-way radio industry since 1997 as well as completing programs in Data Science and Communication Systems to extend my MSEE degree. In my most recent position, I performed radiated cell phone tests and wrote an R language program that parsed memory test data and generated a report with tables of statistical hypothesis tests and graphs. In another case, the report was an interactive webpage where users could zoom-in on data. While my machine-learning and data science work have predicted electric power reliability and cancer health conditions, modeled SQL and MongoDB databases, and built statistical data products, I also build RF engineering dashboards and calculators such as RF power and noise budget tools for the receiver block chain such as the link in the RF section below.
You can see my resume here.
Communications and RF Electronics
My communication work includes radio architecture, antennas, microwaves, noise figure and gain for receiver block chains, low noise amplifiers, digital signal processing and filter design, analog-to-digital converters, and error correction coding.
Engineering Design Projects — Reproducible Performance
I use a reproducible method for engineering design projects that is similar to the data science method. The idea is that a single computer script accomplishes everything from the design calculations to analyzing the data and generating graphs and visualizations in an automated report, nothing should be done manually. Even to change the report, the changes should be made to the script as this insures reproducibility. One way that engineering projects are a little different is that they usually require extra process steps of initial design calculations, and expressing a model such as a netlist or IC layout as a text file in order to generate the raw data for the analysis using a simulation tool such as SPICE, although sometimes I use general linear equation tools in Python or R to achieve the same result.
After the simulated data is generated, Python and R toolsets of data science organize power dissipation levels, frequency response and Fourier analysis, and reliability data in data frames and tables for analysis.
In the server testing environment on my own PC, I can use either the SPICE or the data science tools to solve the equations. For example, the following link describes a tool that designs an RF transformer to match components in an RF chain such as a power amplifier and an antenna. I have two versions of the tool; one calls a SPICE program to solve the equations one the other version solves the equations with linear equation tools in R. I'm investigating the feasibility of installing a SPICE program in the cloud droplet for this website.
I have written LabVIEW programs for automated control and electronic measurements since 2014. I also write user instructions and guidelines for these programs. At Motorola Mobile, I made a Windows LabVIEW application for cellphone battery charging control and measurements. An emulated battery supply was swept through a range of voltage levels as the current was measured in a repeating process loop.
To provide flexibility, the LabVIEW program began by importing a row of data from a spreadsheet file that determined voltages, charging and charger-off times for current measurements, random number seed values for some timing settings, and quality control limits. Serial numbers of equipment and test samples were polled and recorded. Measurements were repeated during each process loop and compared with requirements as key graphs and results were displayed continuously by the Windows program. Finally, all results and measurements were stored in a spreadsheet data file in an open file format for easy analysis with Excel spreadsheets and data science analysis and automated reports. I have also written PLC structured data programs and communication with Beckhoff Twincat-3 software. Serial Communication via RS-232 and RS-485 with Modbus Protocols I design test fixtures with mechanical and electronic capability, Python and R programs to mine test data, create and manage databases, perform statistical analysis and generate automated reports. I am a Registered Professional Engineer in Illinois.
I have performed radiated LTE tests for RF cellular products, measurements with triggered oscilloscopes, and performed micro-soldering to 0.3-mm pads.
I write data science R programs and bash scripts that mine test data, stored it in a data file, perform statistical hypothesis tests, and generate reports with tables and graphs.
I have made interactive dashboards for independent user analysis and plot adjustments.
This webpage uses the nginx server in a cloud droplet at DigitalOcean. I perform testing using my computer at home using the Apache2 server running in Linux Ubuntu as provided by the Windows 10 development tools from Microsoft. I have not found any differences between the Ubuntu running in the cloud and the Ubuntu running in the Linux partition on my PC at home. I also have encountered any problems moving from the Apache2 server at home to the nginx one at the actual website.
I had read about how Dean Attali had created his own shiny server using the nginx server at DigitalOcean. I initially sought to have my own interactive websites here too as I have a number of them already running on the RStudio shiny server. However, since I just completed the University of Michigan Web-Development Certificate by Coursera, this website is just a traditional one with an installed MySQL database based on the website for the final project, but it is changed a little so that it cannot be directly used for the course.
I installed the R program in the cloud droplet for the website. Perhaps the next step might be to install the RStudio shiny server. However, I don't want to risk messing up the current website with such an experiment. I may try it on a new droplet. Since, the R program is running in the cloud droplet, I need to try generating some graphs using R programs called by the nginx server too.