University of Turin HPC researcher focused on federated and distributed learning systems and scalable data parallelism.
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-02) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
Not yet scored: this axis measures publication volume and productivity over time, such as paper counts and output trajectory. The snapshot covers only two reviewed 2026 papers with no aggregate output statistics, so no score can be computed.
Impact—
Not yet scored: this axis would measure citation-based influence, such as citation counts and h-index. No bibliometric impact data is present in this snapshot, so the axis is left unscored rather than estimated.
Focus0/100
Scores 0 on focus, derived from the field distribution: 50% high-performance computing / distributed machine learning systems and 50% distributed machine learning / federated learning systems. The split is even across two fields, though in practice they are adjacent strands of one systems agenda, so the low score partly reflects how finely the catalog classifies this work.
Collaboration32/100
Scores 32 on collaboration: the snapshot records only 4 distinct coauthors across the reviewed papers, a small, concentrated network. Gianluca Mittone recurs on both 2026 papers and Robert Birke is the other frequent partner, suggesting a stable core team rather than broad external collaboration; both papers list Aldinucci as last author without corresponding-author designation.
Activity—
Not yet scored: this axis measures research activity cadence, such as publication frequency and recency across successive years. Only two papers, both from 2026, are available, which is too little to estimate a cadence.
Reproducibility60/100
Scores 60, the mean reproducibility readiness across 2 reviewed papers, and the split is stark: the FSDP/HSDP federated training paper rates high, backed by 512-GPU experiments and a cost model, while the FFL framework vision paper rates low, with thin evaluation and mostly future work.
Research summary
Marco Aldinucci is a researcher at the University of Turin whose reviewed work sits at the intersection of high-performance computing and distributed machine learning systems. The Evolve catalog currently holds two 2026 arXiv papers, both with him in the last author position, indicating a senior or supervisory role. The first, Accelerating Sharded Data Parallelism at Scale with Federated Learning, combines federated learning with FSDP and HSDP sharding and reports results at 512-GPU scale together with a cost model; Evolve's review credits the scale but flags that perplexity gains are confounded by unequal effective batch sizes. The second, Efficiently Distributed Federated Learning, is a vision paper for FFL, a C/C++ federated learning framework built on FastFlow that claims 2.5–3.69x speedups over OpenFL across x86, ARM, and RISC-V; the review judges it promising but thin on evaluation. Bibliometrically the profile is narrow and recent: only four distinct coauthors appear, with Gianluca Mittone and Robert Birke as the recurring partners, and the reviewed output splits evenly between the two fields. Reproducibility is mixed at a mean readiness of 60 across two papers, one high and one low. Output, impact, and activity metrics are missing from this snapshot, so the profile should be read as early-stage.
Collaboration pattern
The reviewed work shows a compact collaboration pattern: only 4 distinct coauthors appear, and both 2026 papers list Aldinucci as last author without corresponding-author designation. Gianluca Mittone recurs across both papers and Robert Birke is the other frequent collaborator, pointing to a stable Turin-centered core team bridging HPC and federated learning rather than a wide external network.
Research keywords
Highlights
Federated sharded data parallelism validated at 512-GPU scale
FFL: a native C/C++ federated learning framework vision
Timeline
- 2026Publishes 'Accelerating Sharded Data Parallelism at Scale with Federated Learning' on arXiv, reporting strong 512-GPU results with high reproducibility readiness.
- 2026Releases 'Efficiently Distributed Federated Learning', a vision paper for the C/C++ FFL framework claiming 2.5–3.69x speedups over OpenFL on x86, ARM, and RISC-V.
- 2026Appears as last author on both reviewed papers without corresponding-author designation, consistent with a senior or supervisory role in the collaboration.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Radar scores are computed from the platform's bibliometric snapshot, which here covers only two reviewed 2026 arXiv papers; output, impact, and activity axes lack sufficient data.
- The frequent-coauthor list derives only from the reviewed catalog and may not reflect the researcher's full collaboration history.
- The reproducibility score of 60 is a mean over just 2 papers (one high, one low) and is not a stable estimate.
- Profile is name-resolved from public bibliographic data via ORCID 0000-0001-8788-0829; country was not provided in the source data.