We present an approach for entangling electron spin qubits localized on spatially separated impurity atoms or quantum dots via a multi-electron, two-level quantum dot. The effective exchange interaction mediated by the dot can be understood as the simplest manifestation of RudermanKittel-Kasuya-Yosida exchange, and can be manipulated through gate voltage control of level splittings and tunneling amplitudes within the system. This provides both a high degree of tuneability and a means for realizing high-fidelity two-qubit gates between spatially separated spins, yielding an experimentally accessible method of coupling donor electron spins in silicon via a hybrid impurity-dot system.Single spins in solid-state systems represent versatile candidates for scalable quantum bits (qubits) in quantum information processing architectures [1][2][3][4][5][6]. In many proposals involving single-spin qubits localized on impurity atoms [2,7] and within quantum dots [1,8], two-qubit coupling schemes harness the advantages of tunnelingbased nearest-neighbor exchange interactions: exchange gates are rapid, tunable, and protected against multiple types of noise [9][10][11][12][13]. These features have been demonstrated for electron spins in quantum dots [14][15][16][17], while a similar demonstration for spins localized on impurity atoms such as phosphorus donors in silicon remains an outstanding experimental challenge [6]. Although the exchange interaction originates from the longrange Coulomb interaction, its strength typically decays exponentially with distance [8,18]. Long-range coupling via concatenation of multiple nearest-neighbor interactions is not ideal for coupling spatially separated electron spins, as it sets a low threshold error rate below which fault-tolerant quantum computing is feasible [19,20]. A mechanism for long-range coupling that simultaneously enables scalability and robustness against errors is therefore key to realizing practical spin-based quantum information processing devices.Approaches to implementing long-range interactions typically involve identifying a system that acts as a mediator of the interaction between the qubits, with proposed systems including optical cavities and microwave stripline resonators [21][22][23][24][25][26] [32,33], and multi-electron molecular cores [34]. Recently, long-range coupling of electrons located in the two outer quantum dots of a linear triple dot system has been demonstrated [35,36]. The effective exchange interaction in that system arises from electron cotunneling between the outer dots and exhibits the fourth-order dependence on tunneling amplitudes that is characteristic of superexchange [37], but suffers from a large virtual energy cost from the doubly occupied center dot states. In contrast, a many-electron quantum dot in the center can also couple distant spins via the Ruderman-Kittel-Kasuya-Yosida (RKKY) interaction, with low-energy intermediate states [38], but perhaps at the cost of low fidelity as impurityFermi sea correlations become hard to disentangle...