Toronto Metropolitan University researcher spanning LLM-based program repair, agent memory, and adversarial ML security
2
papers here
1
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
3
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—
Not yet scored: the underlying OpenAlex record lacks complete venue-level publication metadata. This axis would measure publication volume, venue quality, and authorship roles; yearly counts (2, 2, 5, 8 for 2023-2026) hint at rising volume but do not constitute a scored output measure.
Impact—
Not yet scored: citation coverage is too sparse for a reliable impact estimate; the record lists only h-index 3 and i10-index 1. This axis would measure citation influence, uptake of the dependency-repair and agent-memory work, and standing in the field.
Focus100/100
All field-tagged cataloged work maps to a single field, software engineering (AI for SE, dependency management), giving a perfect focus score of 100. Topic counts reinforce a coherent security-and-languages cluster: software engineering research, blockchain applications and security, and adversarial robustness each appear 4 times, with logic/type systems and network security at 3. The two 2026 papers (dependency repair, agent memory) stay within AI-for-SE and LLM systems.
Collaboration39/100
The OpenAlex record shows 6 distinct coauthors, giving a modest collaboration score of 39. Recurring names include Veronica Poweska, Ariana Oyanguren, Jessica Pourleyli, Theo Rusu, and Manar Alalfi. Both Evolve-cataloged papers place the researcher in a middle author position without corresponding-author status, so the profile reflects contributing membership of small teams rather than independent leadership of large consortia.
Activity81/100
Activity scores 81 on the strength of a steeply rising curve: 2 works in 2023, 2 in 2024, 5 in 2025, and 8 in 2026, with 15 works in the last three years. Both cataloged papers are 2026 arXiv releases, and 2026 is the single most productive year on record, quadrupling the 2023-2024 volume.
Reproducibility55/100
Reproducibility scores 55, computed from a single assessed paper (reproPapers = 1, mean = 55); the cataloged dependency-repair paper carries a medium repro-readiness rating. Because the mean rests on one data point, this axis should be read as indicative rather than established, and additional reviewed releases could move the score meaningfully in either direction.
Research summary
Sourena Khanzadeh is a researcher at Toronto Metropolitan University in Canada whose OpenAlex-resolved profile centers on software engineering, specifically AI for software engineering and dependency management. The record shows a fast-rising but early-career output: yearly work counts grow from 2 in 2023 and 2 in 2024 to 5 in 2025 and 8 in 2026, totaling 15 works in the last three years, with an h-index of 3 and i10-index of 1. The Evolve catalog includes two 2026 arXiv papers. The first, Escaping Python Dependency Hell, presents a hybrid replay-and-repair pipeline in which the PLLM+ configuration lifts HG2.9K dependency repair to 51.9% from a 40.4% baseline, although Evolve notes that 99.7% of fixes replay prior baseline solutions. The second, Selective Forgetting, proposes a graph-based memory framework for long-term LLM agents. Topic counts place software engineering research, blockchain applications and security, and adversarial robustness in machine learning at 4 mentions each, alongside logic and type systems and network security at 3. Six distinct coauthors appear in the record, and both cataloged papers list the researcher in a middle author position, yielding a collaboration score of 39 against a focus score of 100 and an activity score of 81. With reproducibility scored at 55 from a single reviewed paper, parts of this profile remain early-stage and should be read as provisional.
Collaboration pattern
Collaboration is small-scale and stable: 6 distinct coauthors appear in the OpenAlex record, with Veronica Poweska, Ariana Oyanguren, Jessica Pourleyli, Theo Rusu, and Manar Alalfi recurring most often. On both Evolve-cataloged 2026 papers the researcher sits in a middle author position and is not the corresponding author, consistent with a contributing role in compact teams rather than leading large groups.
Research keywords
Highlights
PLLM+ lifts HG2.9K dependency repair to 51.9%
99.7% of generated fixes replay prior baseline solutions
Graph-based memory framework for long-term LLM agents
The 2026 arXiv paper Selective Forgetting proposes a graph-based memory framework with selective forgetting for long-term LLM agents; it is present in the Evolve catalog but has not yet received a full review, so no verdict or repro-readiness rating exists.
Timeline
- 2023First works appear in the OpenAlex record, with 2 publications recorded in the year.
- 2024Output holds steady at 2 works, maintaining a low early-career publication cadence.
- 2025Activity picks up: 5 works are recorded, more than doubling the output of the two preceding years.
- 2026Most productive year on record: 8 works, including two arXiv papers now in the Evolve catalog, the Python dependency repair pipeline and the Selective Forgetting agent-memory framework.
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.
- Bibliometrics are drawn from OpenAlex only; identity is name-resolved from public bibliographic data (no ORCID provided), so homonyms cannot be fully excluded.
- The output and impact axes are unscored (dimMissing); citation coverage beyond h-index 3 and i10-index 1 is not available in this snapshot.
- The reproducibility mean (55) rests on a single assessed paper, so it is indicative only.
- Only one of the two catalog papers carries a full Evolve review; Selective Forgetting lacks verdict, keywords, and repro-readiness metadata.