Machine learning researcher advancing time series forecasting, foundation models, and anomaly detection
3
papers here
3
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-15) 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—
The output axis is not scored yet: no verified publication counts, venue list, or yearly output totals are available for this profile, so this axis cannot currently measure how much peer-reviewed work the researcher produces per year.
Impact—
Impact cannot be scored yet: the profile has no citation counts, h-index, or downstream-usage signals, so this axis cannot currently gauge how much the community builds on these three 2026 arXiv preprints.
Focus0/100
Focus scores 0 on the 0-100 radar, and the raw field distribution explains why: the cataloged output splits evenly across machine learning for time series forecasting, time series anomaly detection, and time series foundation models, each at one third. Rather than a single deep lane, the current portfolio spreads across three adjacent time series subfields at once, which the scoring treats as maximal dispersion rather than concentration.
Collaboration51/100
Collaboration scores 51 out of 100, based on 12 distinct coauthors across the cataloged work. All three 2026 papers are team efforts: he is first author only on the anomaly detection paper and a middle author on both foundation-model papers, indicating a contributor role inside larger research groups rather than consistent solo leadership of projects.
Activity—
Activity is unscored for now: with no multi-year publication calendar beyond the three 2026 arXiv papers, this axis cannot yet measure recent research tempo, submission cadence, or year-over-year consistency of output.
Reproducibility55/100
Reproducibility averages 55 across the three reviewed papers, each rated medium readiness. The Tabby open recipe is singled out as unusually transparent and well ablated, the LLM feature-pool paper's result is judged convincing on TSB-AD-U, while FlowTSFM has pending ablations and a confounded CosMean baseline, so evidential confidence is uneven across the three outputs.
Research summary
Youssef Attia El Hili is a machine learning researcher working across the time series pipeline — anomaly detection, foundation-model pretraining, and probabilistic forecasting. The Evolve catalog holds three of his 2026 arXiv papers, spanning three subfields that each account for a third of his field distribution. On the anomaly detection side, he leads (first author) work on LLM-generated feature pools: a compact 10-feature median/MAD pipeline that reportedly beats neural and statistical baselines on TSB-AD-U, with LLM-proposed features lifting it to 0.588 — a verdict Evolve found convincing. He contributes to two time series foundation model efforts: Tabby, an unusually transparent open pretraining recipe for a 145M-parameter model strong on in-distribution GIFT-Eval, and FlowTSFM, a 38.8M depth-as-transport model whose evidence Evolve flags as preliminary pending ablations. Bibliometrically, the profile is early-stage: output, impact, and activity axes lack sufficient data, collaboration scores 51 with 12 distinct coauthors, and reproducibility scores average 55 across the three papers, each rated medium readiness. All outputs are arXiv preprints, so venue vetting is still ahead.
Collaboration pattern
The cataloged work is distinctly collaborative: 12 distinct coauthors appear across three papers, giving a collaboration score of 51. Recurring partners include Malik Tiomoko, Bahaeddine Abdessalem, Shifeng Xie, Ambroise Odonnat, and Keli Zhang, alongside Zehao Xiao, Jianfeng Zhang, Lujia Pan, Corinne Ancourt, Zhiwei Dong, Lei Zan, and Themis Palpanas. He takes the first-author lead once (anomaly detection) and middle positions on the two foundation-model papers, suggesting team-based research groups rather than solo work.
Research keywords
Highlights
LLM-generated feature pools for time series anomaly detection
Tabby: an open pretraining recipe for time series foundation models
FlowTSFM: turning encoder depth into quantile transport
Timeline
- 2026Posts first-author arXiv paper on LLM-generated feature pools for time series anomaly detection
- 2026Co-authors Tabby, an open pretraining recipe yielding a 145M-parameter time series foundation model
- 2026Co-authors FlowTSFM, a 38.8M-parameter model turning encoder depth into quantile transport
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.
- Bibliometric signals (12 distinct coauthors, field distribution, radar scores) come from OpenAlex-derived public bibliographic data; no institution, country, or ORCID was available to verify identity.
- Output, impact, and activity axes are unscored because citation counts and multi-year publication history are absent from the data.
- All three cataloged papers are 2026 arXiv preprints; peer-review outcomes and venues are not yet reflected.
- The profile is name-resolved from public bibliographic data and covers only the papers Evolve has reviewed.