-
Toolingvnvspec, Machine-Readable Specifications for AI Systems
vnvspec is an open-source Python library from our lab for writing machine-readable V&V specifications that map directly to ISO/PAS 8800, ISO 21448, UL 4600, the EU AI Act, and NIST...
-
Academic PracticeBuild Your Academic Website This Afternoon
Your research is only as visible as your profile. Here is why that matters for senior PhD students and postdocs, and a copy-paste tutorial to get a personal website live...
-
Responsible AIWhat the AI Index 2026 Tells Us About the Gap Between AI Capability and AI Assurance
Stanford HAI's 2026 AI Index confirms what the V&V community has been watching for years. Capability is accelerating, incidents are rising sharply, and the infrastructure for assurance is not keeping...
-
supply chainCatalyzing Supply Chain Research with Open Benchmarks and deepbullwhip
Our lab released deepbullwhip, an open-source Python package that combines a vectorized multi-echelon simulator with a registry-based benchmarking framework. A 155-fold gap between synthetic and real semiconductor demand shows why...
-
AI SafetyCan We Verify LLM Safety Through Hidden-State Geometry
Prompt injection attacks leave geometric fingerprints in an LLM's hidden-state space. We built a notebook to visualize them, then surveyed the research literature to ask whether this is a viable...
-
OpportunityCalling Future PhDs: Summer Research Camp in Applied Optimization & AI
Apply for the MIT-UF-NEU 2026 Joint Summer Research Camp — a free, fully online 3-month program exploring computational methodologies for smart cities and transportation. Deadline: April 17, 2026.
-
Hands-On TutorialOperationalizing NASA's Product Verification Process in a Jupyter Notebook
The NASA Systems Engineering Handbook describes a rigorous product verification process. We operationalize it step by step, turning a flowchart into executable requirements and tests for a simple AI robot...
-
Opportunity
King Hussein Cancer Award: Young Investigator Award for Arab Researchers
The King Hussein Young Investigator Award recognizes early-career researchers (age 45 and younger) for impactful contributions to cancer research. Open to Arab investigators worldwide. Prize: $20,000 USD.
-
Research FoundationsWhat Systems Engineers Know That AI Developers Are Slowly Learning
NASA has spent sixty years learning how to build systems that work millions of miles from help. Two chapters from their Systems Engineering Handbook speak directly to the challenge of...
-
AI SafetyWhat Neural Networks See That We Don't
A traffic light becomes a acorn. The image looks identical to human eyes, but the neural network sees something completely different. This isn't science fiction — it's a real vulnerability,...
-
Academic WritingWhat I See When a Manuscript Lands on My Desk
Most desk rejects aren't about the research — they're about preparation. A handling editor shares patterns from reviewing submissions, a 30-minute pre-submission checklist, and thoughts on responsible AI use in...
-
Talk SummaryWhy Your AI Needs Scalable Verification and Validation
AI is everywhere, including places where failure isn't funny. A summary of how formal verification and validation methods are making AI systems provably safe, from aircraft collision avoidance to supply...
-
AI SafetyWould You Put Your Kids on an AI-Driven Bus?
Saudi Arabia is betting big on AI across transportation, energy, and healthcare. But as these systems move from pilot programs to daily life, a simple question demands an honest answer....
-
AI SafetyThree Papers That Changed How We Prove AI Won't Fail Us
Testing checks a few thousand scenarios. Verification proves correctness across every possible one. Three papers built this field.
-
Deep DiveFrom Simplex to Reluplex: How We Learned to Verify Neural Networks
Reluplex is one of those papers where the core idea is deeply satisfying once you see it. This post builds the intuition from the ground up.
-
Talk SummaryWhy AI Breaks and What ISE Can Do About It
A cancer detector that drops from 93% to 70% when it moves to a new hospital. A stop sign that becomes a speed limit sign with a few stickers. A...
Blog
Perspectives on AI safety, verification, and trustworthy autonomy.