Microsoft NLP researcher spanning LLM agent memory and conversational AI safety, grounded in years of text-mining work
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
23
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—
Publication-volume data needed to score output — total works, venue types, and author-position share — is not available in this snapshot, so no score is given; the axis would measure aggregate and per-year publication productivity across venues and author roles.
Impact—
Citation-impact data is likewise missing from this snapshot, so the axis is unscored; while an h-index of 23 and an i10-index of 35 are on record, the fuller per-paper citation and influence measures this dimension requires are not yet available.
Focus0/100
The focus score of 0 reflects an almost perfectly two-sided portfolio: the OpenAlex field distribution assigns 0.5 to LLM agent systems / NLP and 0.5 to AI safety / computational social science. The historical topic spread — topic modeling (50), NLP techniques (32), information retrieval and search behavior (24), biomedical text mining (18), machine learning in healthcare (18) — shows breadth accumulated over time rather than drift.
Collaboration51/100
The collaboration score of 51 sits mid-range: the recent window records 12 distinct coauthors, matching the 12 frequent collaborators Evolve resolves, including Zachary Johnson, Somya Chatterjee, Min Chen, and Jessica Lin. Both reviewed 2026 papers place him in a middle author position without a corresponding-author role, consistent with industry-team authorship where contribution is shared rather than solo-led.
Activity79/100
Activity scores 79, the strongest axis. The recent three-year window contains 14 works, and yearly counts show a clear rebound: from a low of 1 in 2022 to 4 in 2024, 5 in 2025, and 5 in 2026, after an earlier peak of 14 in 2017. Current pace is near the researcher's historical high-water mark.
Reproducibility55/100
Reproducibility scores 55, computed as the mean over the 2 papers Evolve has reviewed, both rated medium repro-readiness. The mistreatment audit is noted for transparency (5% raw vs 0.90% adjusted detection) but its judge audits are thin; AIM's MUMBench design is promising yet weakened by synthetic circularity and missing baselines — a mixed, mid-range signal.
Research summary
Sadid A. Hasan is a researcher at Microsoft (United States) working across natural language processing, whose recent output splits evenly between LLM agent systems and AI safety / computational social science — the OpenAlex field distribution assigns 0.5 to each. A bibliometric record spanning 2008–2026 shows an h-index of 23 and an i10-index of 35, with a distinctive shape: a 2017 peak of 14 publications in a single year, a trough of 1 in 2022, and a rebound to 4–5 papers annually in 2024–2026, with 14 works in the most recent three-year window. Topical mass concentrates in topic modeling (50 works), NLP techniques (32), information retrieval and search behavior (24), biomedical text mining and ontologies (18), and machine learning in healthcare (18), suggesting a text-mining foundation now redirected toward modern LLM stacks. Evolve has reviewed two 2026 arXiv papers: a dual-detector audit of user mistreatment across 777K arena chats (5% raw, 0.90% adjusted) and AIM, a privacy-aware memory framework for multi-agent, multi-user LLM systems paired with the MUMBench benchmark. Both list him in a middle author position, and both carry medium reproducibility readiness; the pooled reproducibility signal averages 55 across the 2 reviewed papers. Dimension scores put activity highest at 79, with collaboration at 51, reproducibility at 55, and focus at 0, reflecting the even two-way split.
Collaboration pattern
Sadid A. Hasan's collaboration pattern reads as mid-intensity industry teamwork: 12 distinct recent coauthors, matching the 12 frequent collaborators Evolve resolves, with a collaboration score of 51. In both reviewed 2026 papers he sits in a middle author position without a corresponding role, suggesting contribution within larger research-team efforts rather than solo-led projects. Recurring names include Zachary Johnson, Somya Chatterjee, Min Chen, Jessica Lin, and Sulaiman Vesal.
Research keywords
Highlights
Auditing user-to-AI mistreatment at scale
Privacy-aware memory for multi-user LLM agents
Output rebound after a 2022 trough
Yearly publication counts fall to 1 in 2022, then recover to 4 in 2024, 5 in 2025, and 5 in 2026 — 14 works in the most recent three-year window, approaching the 2017 single-year peak of 14.
Timeline
- 2008First recorded publication year in the OpenAlex record, with 1 work — the start of a record now spanning nearly two decades.
- 2017Career-high output: 14 publications in a single year, the maximum in the provided counts-by-year series.
- 2022Lowest point of the series: 1 publication after 3 in 2021, marking the deepest trough of the past decade.
- 2024Recovery under way: 4 publications, beginning a three-year run of sustained output through 2026.
- 2026Two arXiv papers reviewed by Evolve — the conversational mistreatment audit and the AIM memory framework — both in middle author positions.
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 figures (h-index, i10-index, counts by year, topics) come from an OpenAlex snapshot; output and impact dimension scores are missing from this data pull.
- Only two papers (both 2026 arXiv preprints) carry Evolve review verdicts; earlier career work is summarized bibliometrically but not qualitatively reviewed.
- Reproducibility and collaboration signals rest on a small reviewed base (2 papers, 12 coauthors) and may shift as coverage grows.
- Identity resolved from public bibliographic data (OpenAlex; ORCID 0000-0002-5665-7752); affiliation is as provided and not independently verified.