Contemporary monitoring requirements of gases and liquids for demanding applications such as environmental surveillance, medical diagnostics, food and industrial safety, biopharmaceutical process control, homeland security, and others push the limits of existing detection concepts where we may reach their fundamental performance limits. These and other modern monitoring scenarios demand sensing with higher accuracy, enhanced stability, improved sensitivity, and lower power consumption; often all in unobtrusive formats and at low cost. We are developing new generation of sensors that bridge the gap between existing and contemporary required capabilities. Our sensors utilize radio-frequency and optical detection principles and achieve required performance via system analytics. The system analytics is our methodology to deliver high performance sensing via new sensor design rules that include transducer with several uncorrelated outputs, sensed environment (e.g. sensing film or sensing volume) with diverse intrinsic properties detected by the transducer, and multivariate signal processing algorithms (a.k.a. machine learning). We will illustrate the capabilities of these multivariable (multi-parameter) sensors to quantify individual components in mixtures, reject interferences, and correct for environmental instabilities. Our multivariable sensors when coupled with edge data analytics boost data analytics accuracy and reduce data analytics demands for computing and electrical power. Examples of scenarios where such developed multivariable sensors are important include wearable and remotely deployed sensors, autonomous robotics, and home health. In these and many other scenarios, high-performance advantages of traditional mature instruments are cancelled by application-specific requirements that demand unobtrusive form factors, low or no power consumption, no maintenance, and continuous operation.