Jessica Pourleyli

Toronto Metropolitan University · Canada

Unclaimed · AI-structured profileORCID OpenAlex

Toronto Metropolitan University researcher working on LLM-based program repair and Python software security

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-16) 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 cannot be scored yet: this profile rests on a small OpenAlex-indexed snapshot (four 2026 works) without a consolidated publication record, so cumulative productivity stays unmeasured until a fuller works list is linked.

Impact—

Impact cannot be scored yet: citation-based measures in the current record are effectively zero (h-index 0, i10-index 0), which is too thin to rank and typical of a very early-stage profile.

Focus0/100

Focus scores 0 on the 0-100 concentration scale because the record splits evenly between two fields: software engineering (AI for SE, dependency management) at 50% and software security at 50%. Topic data shows the same breadth, with Software Engineering Research and Software Testing and Debugging Techniques at three works each, plus Web Application Security Vulnerabilities at two. This reads as a deliberate dual-track profile rather than diffuse drift, but measured concentration is objectively low for now.

Collaboration39/100

Collaboration scores 39 on the 0-100 scale, with 6 distinct coauthors recorded. The recurring group includes Veronica Poweska, Ariana Oyanguren, Sourena Khanzadeh, Manar Alalfi, Maitreyee Das Urmi and Glaucia Melo. Both catalog papers are multi-authored arXiv preprints, but she holds no corresponding-author role on either, and team sizes are modest — a small, early-stage network rather than a broad international one.

Activity47/100

Activity scores 47. All indexed output sits in a single year: countsByYear shows 4 works in 2026 with nothing earlier in the snapshot, and recent3yWorks is likewise 4. Both catalog papers appeared on arXiv in 2026, so this reflects a profile that has just gone live rather than a lull; the score is capped by the very short observable window.

Reproducibility55/100

Reproducibility scores 55, computed from Evolve's review of 2 papers (reproMean 55, reproPapers 2). Both carry a medium reproReadiness rating: the dependency-repair paper builds on the public HG2.9K benchmark, though 99.7% of its fixes replay prior baseline solutions, and the security paper reports a full static-to-exploit pipeline whose yield depends on same-model self-confirmation.

Research summary

Jessica Pourleyli is an early-career software engineering researcher at Toronto Metropolitan University (Canada), identified via ORCID 0009-0002-6995-514X and an OpenAlex-indexed bibliometric snapshot. Her indexed record is concentrated entirely in 2026 (4 works) and splits evenly between two tracks: software engineering with an AI-for-SE and dependency-management emphasis (50%) and software security (50%). The Evolve catalog holds two of her arXiv papers. As first author, she contributed 'Beyond Static Guarantees', which measures the gap between static-analysis passes and dynamic exploitability in security-sensitive and LLM-generated Python code, reporting that roughly 1 in 7 statically clean samples can be exploited through an LLM pipeline — a result Evolve finds plausible but tempered by same-model self-confirmation. As a middle author, she contributed 'Escaping Python Dependency Hell', a hybrid replay-and-repair pipeline that lifts HG2.9K dependency repair to 51.9% from 40.4%, though 99.7% of fixes replay prior baseline solutions. Topic data places her in software engineering research and software testing/debugging (three works each) and web application security vulnerabilities (two). Her collaboration footprint is compact — six distinct coauthors including Manar Alalfi and Veronica Poweska — and no corresponding-author role is recorded. Reproducibility scores 55 across the two reviewed papers, both rated medium readiness. Citation-based output and impact axes remain unscored, consistent with a record that has only just become visible.

Collaboration pattern

Co-authorship is compact: 6 distinct collaborators, with Veronica Poweska, Ariana Oyanguren, Sourena Khanzadeh, Manar Alalfi, Maitreyee Das Urmi and Glaucia Melo recurring most often. Both catalog papers are multi-author arXiv preprints — Pourleyli is first author on the security paper and holds a middle position on the dependency-repair paper, with corresponding authorship on neither. The pattern fits a lab-centered working group rather than a wide international network.

Research keywords

Python dependency resolutionPyPI ecosystemLLM-based program repairLLM agent pipelinesStatic analysisSAST false negativesDynamic exploit verificationLLM-generated code securityCWE weakness classificationSoftware testing and debugging

Highlights

Replay-and-repair pipeline lifts HG2.9K dependency repair to 51.9%
Evolve's review of this co-authored paper notes that the PLLM+ hybrid replay-and-repair pipeline raises HG2.9K Python dependency repair from a 40.4% baseline to 51.9%, an improvement that stays honest about how much is genuinely new.
Replay-first design: 99.7% of fixes reuse prior baseline solutions
The same review flags that 99.7% of the pipeline's successful fixes replay prior baseline solutions, making the contribution a well-engineered replay-first pipeline whose novelty and generalization remain open questions.
First-author study: about 1 in 7 statically clean samples is exploitable
Evolve's review of her first-authored security paper reports that roughly 1 in 7 statically clean samples can be exploited through an LLM exploit pipeline, quantifying the static-pass dynamic-fail gap; the yield is plausible but tempered by same-model self-confirmation.

Timeline

2026First-authored 'Beyond Static Guarantees' appears on arXiv, measuring the static-pass dynamic-fail gap in security-sensitive and LLM-generated Python code.
2026Co-authored 'Escaping Python Dependency Hell' appears on arXiv, presenting a hybrid replay-and-repair pipeline for Python dependency resolution.
2026OpenAlex snapshot records 4 works in 2026 (recent3yWorks 4), establishing an early bibliometric footprint at Toronto Metropolitan University.

Papers in the Evolve catalog

Frequent co-authors

Veronica Poweska×1 Ariana Oyanguren×1 Sourena Khanzadeh×1 Manar Alalfi×1 Maitreyee Das Urmi×1 Glaucia Melo×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.

Bibliometric scores are computed from an OpenAlex-indexed snapshot only; h-index and i10-index are recorded as 0 and the citation-based axes (output, impact) are unscored.
Only the 2 papers reviewed by Evolve are in scope, and both are 2026 arXiv preprints, so peer-reviewed venue acceptance is not yet reflected.
Name resolution relies on public bibliographic data (ORCID 0009-0002-6995-514X); the name may be shared by multiple people, so identity confidence is limited at this early stage.
The reproducibility score (55) covers only the 2 reviewed papers; no code-release or artifact-evaluation data was available.