Programming quantum computers is currently difficult due to the semantic gap that exists between the high level insight a quantum algorithm designer aims to express and its low-level efficient implementation in terms of quantum circuits. A promising and much needed direction to bridging this gap is the design of high-level quantum programming languages which lift the level of abstraction and provide for a more expressive, natural and intuitive programming interface, closer to classic programming languages. An important recent development in this space is Silq, a comprehensive high-level language that provides both intuitive semantics, automatic uncomputation and safety guarantees. However, a major roadblock to realizing the promise of high-level quantum programming today is the lack of compilers which take as input a high-level quantum program (e.g., in Silq) and compile it to a low-level representation that runs efficiently on a quantum computer. In this work we take a major step to addressing this challenge by introducing the first compilation framework of Silq programs together with a new high-level quantum intermediate representation, called HQIR. HQIR comes with multiple benefits: it enables high-level optimizations that leverage the available structure in the language, it is easier to reason about than a Silq program, it can be effectively simulated and also compiled to a lower IR, ultimately yielding a representation that can run on a quantum hardware.