Systems-minded Monash researcher building real-time endoscopy foundation models and verifiable ML stacks
2
papers here
2
reviewed by Evolve
0
first-author
0
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—
No publication-count or venue data beyond the two cataloged 2026 preprints is available, so this axis is not scored. It would measure total scholarly output, venue quality, and authorship roles across the full record.
Impact—
Citation counts, h-index, and downstream usage are not available in the input data, so impact is left unscored. This axis would measure how often the endoscopy and ML-stack work is cited, reused, or deployed.
Focus0/100
The focus score is 0/100 because the field distribution is an even 50/50 split between endoscopy computer vision / ML systems engineering and software engineering for ML systems. The two cataloged papers map one-to-one onto these halves — woma in medical image analysis, Numbat in ML systems software — so current evidence shows breadth across two adjacent engineering-driven areas rather than concentration in a single field.
Collaboration14/100
The bibliometric record shows a very compact collaboration footprint: exactly 1 distinct coauthor, Thang Tran, who recurs across both cataloged papers, producing a low collaboration score of 14/100. On both 2026 preprints Dang appears in the last author position without corresponding-author designation, consistent with a small two-person team rather than a large lab network. No broader co-authorship graph can be inferred from the available data.
Activity—
Recent-activity signals such as annual publication cadence are insufficient — both cataloged papers fall in 2026 — so this axis is not scored. It would measure publication pace and sustained engagement over time.
Reproducibility55/100
Reproducibility averages 55/100 across 2 papers, with both rated medium repro-readiness. The woma preprint is pre-registered and candid about misses, yet its shipped weights were selected on the test record; Numbat supplies five-level verified training parity with YOLOv8m (0.4956 vs 0.502 mAP) but rests on a single seed and a closed-source framework — strong evidence discipline with clear, documented gaps.
Research summary
Lan Dang is a researcher at Monash University whose published record in the Evolve catalog sits at the intersection of medical-image computer vision and machine-learning systems engineering. The catalog holds two 2026 arXiv preprints, on both of which Dang appears in the last author position without corresponding-author designation: woma, a pre-registered, real-time foundation model for gastrointestinal endoscopy with fine-tuned variants for polyp detection on colonoscopy and gastroscopy, reported to run near 100 frames per second and outperform a TensorRT baseline while candidly documenting its misses; and Numbat, a self-contained machine-learning stack written in Zig whose training parity with YOLOv8m is verified across five levels (0.4956 vs 0.502 mAP). The field distribution is evenly split (0.5/0.5) between endoscopy computer vision / ML systems engineering and software engineering for ML, and the collaboration footprint is compact — one distinct coauthor, Thang Tran — yielding a low collaboration score of 14. Both papers carry medium repro-readiness and a mean reproducibility score of 55 over 2 papers, reflecting disciplined engineering evidence alongside documented caveats. Output, impact, and activity axes currently lack sufficient underlying data and are deliberately left unscored.
Collaboration pattern
The recorded footprint is a tight two-person pattern: 1 distinct coauthor (Thang Tran) recurs across both 2026 preprints, and Dang holds the last author position on each without corresponding-author status. The collaboration dimension scores 14/100, consistent with a compact duo rather than a multi-institution network; no other co-authorship can be inferred from the available data.
Research keywords
Highlights
Real-time endoscopy foundation model at ~100 fps
Pre-registered evaluation with honest misses
Verified training parity in a Zig ML stack
Timeline
- 2026Posts 'woma', a real-time foundation model for endoscopy with fine-tuned polyp-detection and gastroscopy variants, to arXiv as last author.
- 2026Posts 'Numbat', a self-contained Zig machine-learning stack with five-level-verified training parity against YOLOv8m, to arXiv as last author.
- 2026Both cataloged works appear in the same year across two adjacent fields, each with medium repro-readiness and a mean reproducibility score of 55.
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 scores are computed from OpenAlex-derived data only; no ORCID or country field was available for this record.
- Name resolution: 'Lan Dang' is a relatively common name; this profile is resolved from public bibliographic data and may not uniquely identify the individual.
- The catalog covers only two 2026 arXiv preprints; peer-reviewed venues, citation counts, and career history are outside the current data scope.
- Output, impact, and activity radar axes are unscored because the underlying metrics (publication counts, citations, yearly cadence) are missing.