In the neuromuscular analysis, electromyography (EMG) is typically used to analyze aggregate action potential (AP) signals to detect medical abnormalities, activation levels, and recruitment order, or analyze biomechanics. In our previous work, we compared the performance of these off-the-shelf solutions to research-grade EMG machines and found that due to their rigid electrode placement, low sampling rate, and data transmission medium, they are ill-suited for research use, in which data collection must be robust and accurate. We present XTREMIS: a low-cost and portable EMG platform with a small form factor (55 mm $\times$ 35 mm) that has a sample rate comparable to research-grade EMG machines. Indeed, the experiments on eight subjects have shown that not only does XTREMIS functionally outperform technologies, but also its signal quality is high enough to achieve finger movement classification accuracy similar to research-grade EMG machines, making it a suitable platform for research.