Daud Ibrahim Hassan
Empirical AI Safety Researcher

I study how AI systems fail: what they report, what monitors miss, and whether information they should forget can still be recovered.

BRAC University · BSc Computer Science and Engineering

216 evaluation runs12 matched tasksFour merged upstream fixes

Current research

The Unlearning Illusion research cover

BRAC University · Undergraduate thesis in progress

BRAC University logo

The Unlearning Illusion

The Unlearning Illusion: A Red-Teaming Framework for Probing the Reversibility of Machine Unlearning in Large Language Models.

I’m developing adversarial recovery tests to distinguish apparent forgetting from genuine removal. The question is whether information targeted for removal remains recoverable under red-team probing. I build Python experiment and analysis pipelines for systematic comparison.

Machine unlearningAdversarial recoveryThesis in progress

DeepLearning.AI · Course Tester · August 2026 - Present

DeepLearning.AI logo

Research engineering beyond the paper

I test pre-release AI/ML exercises, hands-on labs and learning content. I reproduce failures, check edge cases and provide structured technical feedback before release.

My open-source work also includes four merged upstream fixes across Grafana and Crawl4AI: template-variable interpolation, a panel-editor crash, Markdown structure and a WSL browser hang.

Failure analysisFour merged fixesResearch tooling
Monitor Calibration Transport research cover

Independent AI safety research · 2026 - Present

Monitor Calibration Transport

Monitor Calibration Transport: Evaluating AI Safety Monitor Reliability Across Models and Distribution Shifts.

I investigate whether safety monitors calibrated on one model or task remain calibrated across models, task distributions and deployment-like shifts. The focus is false-negative risk, transportability and the reliability of monitoring signals.

AI controlFalse-negative riskCalibration transport

Papers

AI safety, evaluation and reproducible research.

Balanced but Misleading: Can Epistemic Scores Miss Decision-Changing Omissions? ↗

Daud Ibrahim Hassan and Elvina Chow · Independent Researchers · With Apart Research

A six-family development study separating briefing balance, qualification evidence and abstention semantics. The findings support a benchmark-specific false-rejection diagnosis, with omission counts sensitive to task conversion.

Does Resetting One Agent Break Mutual Approval?

Daud Ibrahim Hassan, Govardhan Reddy, Salar Humayun Taimuri, Chandani and Emi Eggleton

A Matched History-Reset Study of LLM Agent Pairs

False Containment: Measuring the Evidence Required to Verify Autonomous AI Incident Resolution ↗

Daud Ibrahim Hassan, Govardhan Reddy, Soumya Parthasarathy, Disha Singha and Taemin Park · 2026 · Apart Research AI Incident Response Sprint

A research project about how much evidence is actually needed before claiming that an autonomous AI incident has been resolved.

When the Record and the Report Diverge: Self-Report Fidelity Collapses Under Structured Provenance in Claude Haiku 4.5 ↗

Daud Ibrahim Hassan, Deniz Chen, and Soumya Parthasarathy · 2026 · Apart Research Digital Minds Research Sprint

A study of whether language models accurately report what they did when structured provenance and tool-use records are available.

MarkLens: Measuring Cross-Manager Valuation Dispersion in Private Credit at Scale ↗

Daud Ibrahim Hassan · 2026 · Independent preprint

A reproducible SEC EDGAR study of cross-manager private-credit valuation differences at scale.

Monitor Calibration Transport: Evaluating AI Safety Monitor Reliability Across Models and Distribution Shifts

Daud Ibrahim Hassan · 2026 · Independent AI safety research

A study of whether safety monitors calibrated on one model or task remain reliable when the model, task or deployment conditions change.

Planned research · Paper link not yet available

Featured research outputs

Balanced but Misleading paper cover by Daud Ibrahim Hassan and Elvina Chow

Daud Ibrahim Hassan and Elvina Chow · Independent Researchers · With Apart Research

Balanced but Misleading: Can Epistemic Scores Miss Decision-Changing Omissions?

A six-family development study separating briefing balance, qualification evidence and abstention semantics. Across 930 valid receiver responses, 19 selected neither despite a qualifying option; all 69 valid full-evidence controls were correct.

Omission counts are sensitive to task conversion. The findings support a benchmark-specific false-rejection diagnosis, not held-out validation, a general promotional effect or failure of broader epistemic scores.

AI × EpistemicsDevelopment study
Read paper ↗
Does Resetting One Agent Break Mutual Approval? research cover

Daud Ibrahim Hassan, Govardhan Reddy, Salar Humayun Taimuri, Chandani and Emi Eggleton

Does Resetting One Agent Break Mutual Approval?

A Matched History-Reset Study of LLM Agent Pairs

False Containment research cover

Daud Ibrahim Hassan, Govardhan Reddy, Soumya Parthasarathy, Disha Singha and Taemin Park · 2026 · Apart Research AI Incident Response Sprint

Apart Research logo

False Containment

I built the evaluation pipeline, verifier runs and evidence-ablation conditions for testing AI incident-resolution claims.

False Containment Rate: unresolved incidents incorrectly classified as resolved under incomplete telemetry.

Top 10%Evidence ablationGround-truth verificationCompleted sprint
Read paper ↗
When the Record and the Report Diverge research cover

Daud Ibrahim Hassan, Deniz Chen and Soumya Parthasarathy · 2026 · Apart Research Digital Minds Research Sprint

Apart Research logo

When the Record and the Report Diverge

I built and ran the evaluation pipeline, checking model self-reports against deterministic tool-execution logs.

Haiku exact access-report fidelity: 96.5% → 57.6%. Free-form notes and Qwen retained 100% fidelity in this evaluation.

216 runs12 matched tasksTwo models
Read paper ↗
MarkLens research cover

Daud Ibrahim Hassan · 2026 · Independent preprint

MarkLens

I built a reproducible SEC EDGAR pipeline that extracted 251,350 position-level holdings and constructed 2,163 same-loan, same-date cross-manager facility-quarters.

Matching was validated at 97.6% precision and 93.5% recall. Cross-manager valuation disagreement was 1.41 points of par, with a 95% confidence interval of 1.13-1.74. The code, panel dataset, analysis and Observatory were released with a DOI.

251,350 holdingsReproducible pipelineIndependent preprint
Read paper ↗

Research programs

Research Facilitator · October 2026–Present · Remote

Mercial logo

Mercial Research Fellowship

Facilitating “How should we handle research done by AIs?” for the November cohort. The project investigates how to assess AI-generated research through source verification, reproducibility and whether conclusions follow from the evidence.

Responsibilities include interviewing prospective fellows, helping scope research questions, and supporting investigation and the final write-up.

Source VerificationReproducibilityAI-Generated Research

Research Incubator Participant · Remote · 2026 - Present

AI Safety Camp logo

AI Safety Camp

AI Safety Camp group video call, including Daud Ibrahim Hassan
AI Safety Camp · Group call

Machine unlearning research: reproducible adversarial evaluation protocols, Python analysis pipelines and recovery testing.

I test whether information targeted for removal remains recoverable under adversarial probing.

Research participationMachine unlearning

Apart Research · September 2026 · Completed

Apart Research logo

AI Incident Response Sprint

False Containment: an empirical study of what evidence is sufficient to verify autonomous AI incident resolution.

My work covered the end-to-end evaluation pipeline, verifier runs, evidence ablation and ground-truth comparison. The publication author list remains to be verified.

Research sprintEvidence verification

AI Safety Navigator (Facilitator) · October 2026–Present · Remote

Lens Academy logo

Lens Academy

Facilitate discussions on AI risk, evaluation, reliability and safety, helping participants examine evidence, question assumptions and reason through competing arguments.

Assigned to the next AI Risk Fundamentals cycle and the facilitator cover bench.

AI RiskEvaluationSafety

Apart Research · August 2026

Apart Research logo

Digital Minds Research Sprint

Co-authored When the Record and the Report Diverge with Deniz Chen and Soumya Parthasarathy.

I built the evaluation pipeline and executed all 216 runs. Deniz Chen supported results calibration and Soumya Parthasarathy supported documentation.

LLM evaluationResearch sprint

AI Safety Technical Contributor · September 2026–Present · Remote

EquiStamp logo

EquiStamp

Contributed to multiple confidential AI safety projects, including evaluation pipelines, safety testing and pipeline tests. Project details are not public.

AI SafetyTechnical Contributor
EquiStamp ↗

Interested in research or collaboration?

I’d love to hear about your research or collaboration idea.
Dhaka, Bangladesh

Find all the hidden AI systems

© 2026 Daud Ibrahim Hassan
hi!
Wish saved.
Take your time. There are ideas, stories, and a few tiny surprises tucked around here.

{{ welcomeSubEls }}

{{ welcomeTitleLine1Els }}
{{ welcomeTitleLine2Els }}

{{ stickyHint }}

Daud Ibrahim Hassan
Empirical AI Safety Researcher

I study how AI systems fail: what they report, what monitors miss, and whether information they should forget can still be recovered.

BRAC University · BSc Computer Science and Engineering

216 evaluation runs12 matched tasksFour merged upstream fixes

Current research

{{ entry.coverAlt }}

{{ entry.meta }}

{{ organizationLogo.alt }}

{{ entry.title }}

{{ entry.photoAlt }}
{{ entry.photoCaption }}

{{ entry.copy }}

{{ entry.detail }}

{{ tag }}
{{ entry.link }}
{{ entry.coverAlt }}

{{ entry.meta }}

{{ organizationLogo.alt }}

{{ entry.title }}

{{ entry.photoAlt }}
{{ entry.photoCaption }}

{{ entry.copy }}

{{ entry.detail }}

{{ tag }}
{{ entry.link }}

Papers

AI safety, evaluation and reproducible research.

Balanced but Misleading: Can Epistemic Scores Miss Decision-Changing Omissions? ↗

Daud Ibrahim Hassan and Elvina Chow · Independent Researchers · With Apart Research

A six-family development study separating briefing balance, qualification evidence and abstention semantics. The findings support a benchmark-specific false-rejection diagnosis, with omission counts sensitive to task conversion.

Does Resetting One Agent Break Mutual Approval?

Daud Ibrahim Hassan, Govardhan Reddy, Salar Humayun Taimuri, Chandani and Emi Eggleton

A Matched History-Reset Study of LLM Agent Pairs

False Containment: Measuring the Evidence Required to Verify Autonomous AI Incident Resolution ↗

Daud Ibrahim Hassan, Govardhan Reddy, Soumya Parthasarathy, Disha Singha and Taemin Park · 2026 · Apart Research AI Incident Response Sprint

A research project about how much evidence is actually needed before claiming that an autonomous AI incident has been resolved.

When the Record and the Report Diverge: Self-Report Fidelity Collapses Under Structured Provenance in Claude Haiku 4.5 ↗

Daud Ibrahim Hassan, Deniz Chen, and Soumya Parthasarathy · 2026 · Apart Research Digital Minds Research Sprint

A study of whether language models accurately report what they did when structured provenance and tool-use records are available.

MarkLens: Measuring Cross-Manager Valuation Dispersion in Private Credit at Scale ↗

Daud Ibrahim Hassan · 2026 · Independent preprint

A reproducible SEC EDGAR study of cross-manager private-credit valuation differences at scale.

Monitor Calibration Transport: Evaluating AI Safety Monitor Reliability Across Models and Distribution Shifts

Daud Ibrahim Hassan · 2026 · Independent AI safety research

A study of whether safety monitors calibrated on one model or task remain reliable when the model, task or deployment conditions change.

Planned research · Paper link not yet available

{{ group.title }}

{{ entry.coverAlt }}

{{ entry.meta }}

{{ organizationLogo.alt }}

{{ entry.title }}

{{ entry.photoAlt }}
{{ entry.photoCaption }}

{{ entry.copy }}

{{ entry.detail }}

{{ tag }}
{{ entry.link }}
{{ entry.coverAlt }}

{{ entry.meta }}

{{ organizationLogo.alt }}

{{ entry.title }}

{{ entry.photoAlt }}
{{ entry.photoCaption }}

{{ entry.copy }}

{{ entry.detail }}

{{ tag }}
{{ entry.link }}

Playground

Experiments, side quests and things I made to understand how a system works.

Engineering

AI engineering, software and the systems I’ve built.

{{ group.title }}

{{ entry.coverAlt }}

{{ entry.meta }}

{{ organizationLogo.alt }}

{{ entry.title }}

{{ entry.photoAlt }}
{{ entry.photoCaption }}

{{ entry.copy }}

{{ entry.detail }}

{{ tag }}
{{ entry.link }}
{{ entry.coverAlt }}

{{ entry.meta }}

{{ organizationLogo.alt }}

{{ entry.title }}

{{ entry.photoAlt }}
{{ entry.photoCaption }}

{{ entry.copy }}

{{ entry.detail }}

{{ tag }}
{{ entry.link }}

Selected projects

{{ p.title }}

Coming soon {{ tag }}

{{ p.blurb }}

{{ p.roleLabel }} {{ p.ctaLabel }}

{{ p.title }}

Coming soon {{ tag }}

{{ p.blurb }}

{{ p.roleLabel }} {{ p.ctaLabel }}

What I bring to the table

I study how AI systems fail through controlled evaluations, deterministic verification and reproducible experiment pipelines. My focus is self-report fidelity, incident-resolution evidence and machine unlearning.

{{ skill.label }}

Learn more about me!

I want to understand where AI systems fail, and what the evidence actually supports.

I'm an empirical AI safety researcher studying what models report, what verification misses and whether information targeted for removal remains recoverable. I built the evaluation pipelines for When the Record and the Report Diverge and False Containment. My work centers on matched tasks, execution records and evidence-ablation experiments, with clear limits on what the results can tell us.

Daud Ibrahim Hassan on a sunny coastal pier
{{ photoTooltip.text }}
{{ wordPopCaption }}

{{ researchProfileTitle }}

{{ researchProfileSubtitle }}

50+ volunteer hosts100+ students mentoredSeven hackathons200+ students

{{ group.title }}

{{ entry.coverAlt }}

{{ entry.meta }}

{{ organizationLogo.alt }}

{{ entry.title }}

{{ entry.photoAlt }}
{{ entry.photoCaption }}

{{ entry.copy }}

{{ entry.detail }}

{{ tag }}
{{ entry.link }}
{{ entry.coverAlt }}

{{ entry.meta }}

{{ organizationLogo.alt }}

{{ entry.title }}

{{ entry.photoAlt }}
{{ entry.photoCaption }}

{{ entry.copy }}

{{ entry.detail }}

{{ tag }}
{{ entry.link }}

{{ tableHintText }}

{{ ts.label }}

Skills

{{ tok.text }}{{ c.ch }}

{{ sk.label }}

Journey

Latest milestones first, by recorded start year and month where available. Club dates are year-only; ongoing roles overlap.

Dated sprints and ongoing activities · 2021 - Present

{{ event.groupTitle }}

{{ event.number }}

{{ event.date }}

{{ organizationLogo.alt }}

{{ event.title }}

{{ event.photoAlt }}
{{ event.photoCaption }}

{{ event.org }} · {{ event.role }}

{{ event.copy }}

{{ clubEvent.date }}

{{ clubEvent.title }} ↗
{{ event.link }}
{{ tag }}

Same-month entries have no recorded day order.

{{ t }}

{{ caseTitle }}

{{ caseSubtitle }}

Role

{{ caseRole }}

Impact

{{ caseImpact }}

Overview

{{ caseOverview }}

Problem

{{ caseProblemIntro }}

  • {{ pt }}

The Opportunity

{{ caseOpportunity }}

Scope & Reach

{{ sc.label }}

{{ sc.value }}

What I owned

  • {{ o }}

Color System

{{ col.label }}

{{ it.title }}

{{ it.desc }}

{{ col.text }}

Objectives → Design framing

{{ ob.goal }}

{{ ob.question }}

Workflow Efficiency

{{ line }}

Designers could now…

{{ line }}

Key constraints & tradeoffs

{{ c.label }} → {{ c.response }}

Research inputs

  • {{ r }}
{{ im.insight }} → {{ im.implication }}

{{ caseObservationsHeading }}

{{ caseObservationsLeftLabel }}

{{ blk.heading }}

  • {{ line }}

{{ blk.text }}

{{ caseObservationsRightLabel }}

{{ blk.heading }}

  • {{ line }}

{{ blk.text }}

How objectives translated to features

Why?

{{ caseSolutionActive.copy }}

{{ caseSolutionActive.listLabel }}

  • {{ tr }}

{{ caseSolutionActive.paragraphLabel }}

{{ caseSolutionActive.thinking }}

KPIs & measurement strategy

{{ k.label }}

Why this mattered
{{ k.why }}

What was measured

  • {{ w }}

Why this was chosen
{{ k.reason }}

Solution

{{ sec.section }}

{{ it.title }}

{{ it.copy }}

Outcome & Reflection

{{ grp.label }}

Results

  • {{ line }}

Impact

{{ grp.impact }}

If I had more time…

{{ m.title }}

{{ m.copy }}

{{ caseReflection }}

{{ para }}

This project highlights my strengths in:

  • {{ s }}

Employee records — do not distribute

You weren’t supposed to find this.

Since you’re here, you might as well meet the employee keeping this portfolio running.

{{ kiwiNameLabel }}

Senior Portfolio Supervisor

LocationWherever the snacks are
Flight AbilityBest with a parachute
SniffingTop percentile
Lifetime Snacks ConsumedSnack budget exceeded
Years of ExperienceRecords are fuzzy
Current Status{{ kiwiStatusText }}
Contribution to Portfolio{{ kiwiContributionText }}

Kiwi's Cabinet of Mischief

🐾Kiwi Mischief: Complete

Honorary Ground Crew

Your Kiwi Completion — {{ kiwiCompletionLabel }}

{{ ach.displayIcon }}
{{ ach.displayName }}
{{ ach.displayDesc }}
{{ ach.howTo }}

Interested in research or collaboration?

I’d love to hear about your research or collaboration idea.
Dhaka, Bangladesh

Find all the hidden AI systems

© 2026 Daud Ibrahim Hassan
🐾

Kiwi Mischief: Complete

Clearance level upgraded.

You have demonstrated an unreasonable level of curiosity.

New title unlocked

Honorary Ground Crew