Toqeer Ali Syed

Islamic University of Madinah · Saudi Arabia

Unclaimed · AI-structured profileOpenAlex

Accessibility and federated-learning researcher building edge-AI assistive systems for visually impaired users.

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-17) 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—

Output is unscored: the snapshot lacks the cumulative publication-volume data this axis measures (total works over time by year and type). With only 3 visible 2026 preprints, any lifetime productivity figure would be guesswork until fuller coverage accumulates.

Impact—

Impact is unscored: the citation record this axis needs has not accumulated — h-index and i10-index both read 0, indicating no aggregated citation signal yet rather than a measured zero impact. Scores will follow once citations accrue to the 2026 preprints.

Focus0/100

Focus scores 0 because the field distribution is an exact three-way split — 0.333 each for federated-learning security and privacy, assistive technology/embedded edge AI, and accessibility-oriented HCI with multi-agent systems. Topic tags (tactile and sensory interactions and assistive technology, 2 each; three further topics at 1) confirm breadth across distinct communities. This is an early-career breadth signal, not necessarily a permanent pattern.

Collaboration41/100

Collaboration scores 41. The snapshot records 7 distinct co-authors, with Ali Akarma, Muhammad Khan, Qurat-ul-ain Mastoi, Adeel Ahmad, and Danial Hameed appearing most frequently; Adeel Ahmad is listed three times, suggesting the closest recurring partner. Authorship rotates — first author on the multi-agent paper, last on the smart cane, middle on the federated-learning study — with none marked corresponding, consistent with small-team work.

Activity40/100

Activity scores 40. countsByYear shows all 3 recorded works landing in 2026, and the recent-three-year window likewise counts 3 works, so visible output is compressed into a single recent year rather than spread across a track record. All three items are arXiv preprints. With no earlier per-year history in the snapshot, the moderate score reflects a newly visible publishing cadence rather than sustained multi-year throughput.

Reproducibility32/100

Reproducibility scores 32, the mean across 3 reviewed papers. Evolve rates repro readiness low for the multi-agent orchestration paper (design-stage, unvalidated) and the smart cane (no baselines, ablation, or visually impaired-user validation despite reporting F1 0.82 at 330 ms on an $88 build), and medium for the federated-learning study (solid clip-then-noise work, but an internally inconsistent DP budget table). Released artifacts and user validation are the gaps.

Research summary

Toqeer Ali Syed is a researcher at the Islamic University of Madinah (Saudi Arabia) whose visible work spans three adjacent areas: accessibility and assistive AI for blind and visually impaired users, embedded edge AI, and machine-learning security, specifically federated-learning privacy. The Evolve catalog holds three 2026 arXiv preprints. As first author, Syed proposes an eight-agent orchestration architecture giving non-visual access to virtual worlds; Evolve judges it coherent but design-stage and unvalidated. As last author, Syed contributed an $88 offline smart cane on a Raspberry Pi Zero 2W reporting F1 0.82 and 330 ms latency — credible engineering with honest limits, but no baselines, ablations, or visually impaired-user validation. In a middle-author position, Syed co-authored a gradient-leakage study claiming about 100% client de-anonymization from federated updates on a HAR proxy, with a clip-then-noise differential-privacy defense whose budget table Evolve flags as internally inconsistent. OpenAlex coverage is early-stage: 3 works in the recent three-year window (all 2026), h-index and i10-index of 0, 7 distinct co-authors, and an exact one-third field split across the three areas, which yields a focus score of 0. Available dimension scores are collaboration 41, activity 40, and reproducibility 32 (a mean over the three reviewed papers), while output and impact remain unscored for insufficient data. Overall, this is a newly visible profile whose emphasis is assistive, privacy-preserving AI for users at the margins.

Collaboration pattern

The snapshot records 7 distinct co-authors, indicating small-team work rather than large consortia. The most frequent partners are Ali Akarma, Muhammad Khan, Qurat-ul-ain Mastoi, Danial Hameed, and Adeel Ahmad — the last listed three times, pointing to the closest recurring collaboration. Authorship roles rotate across first, middle, and last positions, with no paper marked corresponding, consistent with an early-stage researcher contributing at different levels across parallel assistive-AI and federated-learning projects.

Research keywords

accessibilityassistive technologyblind and visually impaired userssmart caneedge AIobject detectionmulti-agent systemsfederated learninggradient leakageclient identity inferencedifferential privacyvehicular edge networks

Highlights

Multi-agent orchestration for non-visual access to virtual worlds
First-author 2026 arXiv preprint proposes 8 specialized agents plus an orchestrator arbitrating non-visual output for blind and visually impaired users in virtual worlds; Evolve's verdict: coherent but design-stage and unvalidated.
An $88 offline smart cane for multimodal mobility assistance
Last-author 2026 preprint presents an $88 offline smart cane on a Raspberry Pi Zero 2W with TensorFlow Lite object detection, reporting F1 0.82 and 330 ms latency; credible engineering with honest limits, but no baselines, ablation, or VI-user validation.
Gradient leakage exposes federated-learning clients in vehicular edge networks
Co-authored 2026 study claims about 100% client de-anonymization from gradient-based identity inference on federated updates, using a HAR proxy for inertial sensing; the clip-then-noise defense study is solid, though Evolve flags an internally inconsistent DP budget table.

Timeline

2026Co-authors the federated-learning gradient-leakage study on inertial sensing in vehicular edge networks, claiming about 100% client de-anonymization with a clip-then-noise defense.
2026Publishes a first-author arXiv preprint on eight-agent orchestration for non-visual access to virtual worlds (Evolve verdict: design-stage, unvalidated).
2026Publishes as last author the $88 offline Raspberry Pi Zero 2W smart cane reporting F1 0.82 and 330 ms for multimodal mobility assistance.

Papers in the Evolve catalog

Frequent co-authors

Ali Akarma×3 Muhammad Khan×1 Qurat-ul-ain Mastoi×1 Adeel Ahmad×1 Adeel Ahmad×1 Adeel Ahmad×1 Danial Hameed×1

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.

Bibliometrics are from OpenAlex only; no ORCID is linked, so identity is name-resolved from public bibliographic data and may merge or split same-name authors.
Citation impact has not accumulated: h-index and i10-index are recorded as 0, so the output and impact dimensions are left unscored.
All three reviewed papers are 2026 arXiv preprints; peer-reviewed venue publication status is unknown.
The reproducibility score is the mean over only 3 reviewed papers (readiness: low, low, medium) and may not reflect the full research portfolio.