The terahertz (THz) spectral range is devoid of commercially feasible radiation sources, detectors, and components. In particular, THz sources are bulky, complex to operate, and cost-prohibitive - more suited for a research laboratory than a commercial setting. Developing compact and mass-producible sources in the 1 to 6 THz spectral range will open up new avenues for this technology to make a mainstream societal impact. The focus of this thesis is the development of compact, room-temperature terahertz sources based on quantum cascade lasers (QCL) and quantum well technology. QCLs are semiconductor lasers that operate with high power at mid-infrared (mid-IR) and THz frequencies. THz QCLs are the only mW-level average power sources with spectral coverage from 0.8 to 5 THz. However they only work at cryogenic temperatures because they cannot maintain population inversion across the lasing transition at elevated temperatures. Cryogenic cooling makes these sources cumbersome to operate and expensive to manufacture. Room-temperature operation significantly enhances their commercial appeal and a portion of this dissertation investigated the improvement in THz QCL temperature performance using GaAs-Al₀.₁₅Ga₀.₈₅As double-phonon resonant active region designs. These devices worked up to 173 K and were a substantial improvement compared to prior implementations of double-phonon resonant designs. Room-temperature THz sources that do not require population inversion across the lasing transition can be engineered by combining the field of nonlinear optics with intersubband transitions in quantum well structures. One method of creating inversionless THz lasing is based upon the principle of Raman gain in semiconductors and this thesis explores the design of an intersubband Raman laser (IRL) with GaAs-Al₀.₃₃Ga₀.₆₇As heterostructures. The primary focus of this dissertation is developing room-temperature, broadly-tunable, monolithic THz sources based on difference-frequency generation (DFG) in mid-IR QCLs. The source active region is quantum-engineered to provide lasing at mid-IR frequencies, ω₁ and ω₂, and simultaneously have giant second-order optical nonlinearity for THz generation at frequency ω [subscript THz]=ω₁–ω₂. This dissertation developed a Cherenkov emission scheme that produced devices with a narrow emission linewidth, 0.12 mW peak power and tuning from 1.55 to 5.7 THz - the largest tuning bandwidth compared to semiconductor technology of similar size and cost.