UIC researcher: runtime oversight of self-modifying LLM agents and reliability-aware gait measurement for rehabilitation
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-13) 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 scorable: publication counts, venue types, and cumulative volume are absent from the current snapshot. This axis would measure total indexed output and its distribution across venues and years.
Impact—
Not yet scorable: citation counts are unavailable in the provided data. This axis would measure citations to the reviewed papers and field-normalized influence once uptake data is indexed.
Focus0/100
The focus score is 0, computed from a field distribution split exactly 50/50 between artificial intelligence (LLM agents and oversight) and computer vision for multimodal human motion capture in rehabilitation. Each of the two cataloged 2026 papers sits in a different line, so the portfolio reads as a deliberately dual-track agenda rather than a concentrated one.
Collaboration39/100
Collaboration scores 39 from 6 distinct coauthors across the cataloged work, a compact footprint. The recurring co-author list — Sina Tayebati, Divake Kumar, Nastaran Darabi, Ranganath Krishnan, Nethmi Jayasinghe, and Mihir Parashar — suggests a stable core team, with Trivedi taking the senior last-author position on both reviewed 2026 papers.
Activity—
Not yet scorable: only 2026 papers are cataloged, so no multi-year cadence exists. This axis would measure recent publication frequency and year-over-year continuity once more indexed years are available.
Reproducibility55/100
Reproducibility scores 55, the mean over the 2 papers Evolve assessed, each rated medium readiness. The reviews flag concrete gaps — the agent-oversight paper reports only 2/16 confidence intervals excluding zero and omits a key gating ablation, while GaitVista delivers strong controlled evidence but tests healthy subjects only.
Research summary
Amit Ranjan Trivedi is a researcher at the University of Illinois Chicago whose reviewed work, as indexed by Evolve, spans two equally weighted lines — artificial intelligence (LLM agents and oversight) and computer vision for multimodal human motion capture in rehabilitation — each accounting for half of the catalog's field distribution. Both cataloged papers are 2026 arXiv releases on which he holds the last (senior) author position. The first, Self-Healing Harness for Runtime Oversight of Agent Self-Modification, proposes evidence-gated admission for agent self-modification; Evolve's review credits a strong motivation (211/383 regressions) but flags that only 2/16 confidence intervals exclude zero and that a key gating ablation is missing. The second, GaitVista, introduces a reliability gate for visual-inertial gait fusion that cuts TotalCapture error by 27.7%, with strong controlled evidence, though healthy-only subjects limit its accessibility claims. Bibliometrics show a moderate collaboration footprint (score 39; 6 distinct coauthors) and a reproducibility mean of 55 across 2 papers rated medium readiness, while the focus score of 0 reflects the even split between two distinct fields. Output, impact, and activity metrics are not yet available in this snapshot.
Collaboration pattern
Trivedi's cataloged work shows a compact collaboration footprint: 6 distinct coauthors and a collaboration score of 39. The same six names recur — Sina Tayebati, Divake Kumar, Nastaran Darabi, Ranganath Krishnan, Nethmi Jayasinghe, and Mihir Parashar — pointing to a stable core group rather than wide external networking. On both reviewed 2026 papers he sits in the last (senior) author position, with no corresponding-author flag recorded, consistent with supervising two parallel project lines.
Research keywords
Highlights
Evidence-gated admission for self-modifying LLM agents
Reliability gate cuts vision-IMU gait error by 27.7%
Timeline
- 2026Posts Self-Healing Harness for Runtime Oversight of Agent Self-Modification to arXiv as last author, opening the LLM-agent oversight line.
- 2026Posts GaitVista on reliability-aware longitudinal gait assessment to arXiv as last author, anchoring the rehabilitation-vision line.
- 2026Both cataloged papers land in the same year, splitting the catalog's field distribution evenly between AI oversight and rehabilitation motion capture.
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 indicators come from OpenAlex-indexed records only; output, impact, and activity scores are unavailable in this snapshot.
- Coverage is limited to two 2026 arXiv papers reviewed by Evolve and may not reflect the full publication record.
- The reproducibility score averages only 2 assessed papers, both rated medium readiness — treat as early-stage evidence.
- Author-position detail is limited to recorded catalog fields; this profile is name-resolved from public bibliographic data.