This paper proposes a wearable device, CigEst, to estimate the number of cigarettes the wearer has smoked and alerts the wearer using haptic feedback and a mobile phone application. The requirement for such a device is obvious in the backdrop of a staggering $1.134 trillion cigarette market, and increasing health and financial risks associated with smoking. The wearable device presented in this paper, CigEst, is a proof of concept built around a Bluetooth-enabled microcontroller integrated with a 6-axis accelerometer and a carbon monoxide gas sensor. The movement of the hand when the wearer smokes, while not unique, is a repetitive pattern of bringing the hand to the mouth. This movement can be detected using an accelerometer. A dataset was created by having multiple wearers wear the device while smoking. A machine learning model was built using this accelerometer data, and is deployed on the wearable device by reducing the model. The carbon monoxide sensor is used to trigger the machine learning model, since the detected hand movement is not unique to smoking. The number of hand movements to the mouth is recorded and the number of cigarettes smoked is estimated using this data. The same is sent through a mobile phone to an IoT server for long term storage and further analysis. The smoking action detection algorithm shows a maximum of 95.5% accuracy in detecting a smoking action, and 88% on deployment of the ML model on the wearable device.