Human errors, unavailability of patient history, lack of demographic data, lack of expert supervisors for time sensitive diagnosis etc. often lead to fatal errors in Cardiovascular Disease diagnostics (CVD). In our current work, we incorporated machine learning through deep convolutional neural network and wireless bio-sensing to make an ultraportable ECG (Electro Cardiogram) module that will help reduce these errors. We’ve verified our system and algorithm using MIT-BIH and PTB database of 40 patients with an accuracy of 84%. As we trained our system with more patients, our accuracy improved from 58% for 10 patients to 84% for 80 patients. The preliminary system showed 92.3% accuracy with 7.69% false positives. Our wireless sensor nodes utilized low power/ low bandwidth MQTT protocol to mimic a very low latency wired equivalent real-time system with ultra-portability.