Assistant Professor
Department of Computer Science
VU Amsterdam
d.bonetta@vu.nl · Scholar · DBLP · GitHub · LinkedIn
I work at the intersection of programming languages and data systems. I am particularly interested in programming language implementation technologies (e.g., compilers, language VMs, optimizations), and how they meet modern data systems in areas such as data formats, parsing, querying and hardware acceleration. The recurring theme of my research is bridging the gap between high-level programming abstractions and the systems and hardware that execute them.
Before joining the VU I was a Principal Researcher in the Virtual Machine Research Group at Oracle Labs and a member of the GraalVM team, where I also served on ECMA TC39. I hold a Ph.D. from USI Lugano. At the VU I am part of the Computer Systems and At Large team. I also hold a part-time appointment in the Database group of TU/e.
Recent news
- OOPSLA 2026TwinString: Preserving string semantics with off-heap data on the JVM. Keeping Java’s String semantics intact while the characters live off the heap.
- ICSME 2026Unavailable, Undocumented, or Unbuildable: A Reproducibility Study of WebAssembly Benchmarks. How many published WebAssembly benchmarks can still be obtained, understood and built.
- OOPSLA 2025HeapBuffers: Why Not Just Using a Binary Serialization Format for Your Managed Memory? Using a binary serialization format as the layout of managed memory itself.
- VLDB 2025GpJSON: High-performance JSON Data Processing on GPUs. JSON parsing and querying executed on the GPU.
- CF 2025Multi-GPU Greedy Scheduling Through a Polyglot Runtime. Scheduling work across several GPUs from inside a language runtime.
- IPDPSW 2025Memory Efficient WebAssembly Containers. Cutting the memory footprint of WebAssembly-based containers.
Ongoing work
VMs, Runtimes and compilers
JIT and AOT compilation in JVM and WebAssembly runtimes: profile-guided inlining, stateful incremental compilation, SIMD-aware optimizations.
Data formats and memory layout
Serialization, compressed representations and managed-heap layouts — Arrow, Parquet, JSON and friends — and the cost they impose on the runtime.
Accelerators for data processing
Moving parsing, joins and aggregation onto GPUs and NPUs, and finding out which parts of a query engine actually survive the trip.
Cloud workloads
Startup latency, memory footprint and energy use of Java/GraalVM and WebAssembly in serverless and container settings.
Teaching
- Systems Seminar M.Sc. — VU Amsterdam
- Language VMs: Design and Implementation M.Sc. — TU Eindhoven
- Advanced Systems Programming B.Sc. — VU Amsterdam
Working with me
I am always looking for students who like building things: practical tools, careful experimental evaluations, real contributions to open-source projects. I supervise Ph.D. students at the VU, and B.Sc. and M.Sc. theses, literature studies and research projects at both the VU and TU/e — the current open topics are a good place to start.
Contact
d.bonetta@vu.nl (VU)
d.bonetta@tue.nl (TU/e)
Office
VU Amsterdam, Faculty of Science
NU building, 11A-43
De Boelelaan 1111, 1081 HV Amsterdam