Latest
The complete chronological feed across writing, projects, talks, and judging. Most recent first.
143 entries
Evidence first, arithmetic second: a system report and failure analysis for DocSem
Shared-task system paper at DocInsights 2026, co-located with EMNLP 2026, on reading a figure out of a PDF before computing with it.
Program Committee member for the IEEE BigData 2026 High School Symposium. Reviewed three student papers against a standard set for high school research.
Test before release. Enforce at runtime.
Talk on governance for tool-using AI agents: why a correct answer can still be unsafe, the three checks that catch it in CI, and the runtime gate that stops it before the action runs.
Reviewer for IEEE Access, IEEE's multidisciplinary open-access journal. Reviewed one submitted manuscript.
Artifact Evaluation Committee member for ACSAC 2026, the 42nd IEEE Annual Computer Security Applications Conference. Evaluated two artifacts for reproducibility.
Follow the pointer. Then check the passage.
Review a fictional answer at Sachin. Follow its citations, repair missing support, and inspect how citation recall and precision are calculated. An illustrated review desk and a runnable JavaScript example connect the story to the mechanism.
Reviewer for Global South AI: Rethinking AI for and from the Global South at NeurIPS 2026. Reviewed three submitted papers.
Follow one request from the badge desk to the response.
Run a local permission-aware answer service. Test forged group claims, tenant isolation, revoked access, stale caches, and source requests.
Reviewer for the Third Workshop on Agents in the Wild: Safety, Security, and Beyond at NeurIPS 2026 (Sydney). Reviewed three submitted papers.
The envelope matters as much as the letter.
Inspect four places information can escape an otherwise filtered answer: citations, previews, shared answers, and diagnostics.
