Corey Alejandro · AI Engineer • Instructional Systems Designer • Researcher • Maker
Stanford University alumnus (B.A., Drama). Currently enrolled in the MS in AI Engineering program at Quantic School of Business and Technology.
I build ways for people to understand complex systems, make informed decisions, and keep authority over what happens next. Instructional systems design is the established through-line connecting my AI engineering, research, learning experiences, dramaturgy and adaptive-fashion work.
Work in Motion — come see how I think
- Contract Window for Bob: intent, authorized action and inspectable evidence in AI-assisted engineering. An implemented hackathon submission; the illustrated IDE interface is not a claim of a validated runtime integration.
- Reflexive: source-preserving, contestable research workflows. Prototype and Catalyst grant application; no award or validated pilot is claimed.
- MADE HERE / MOVES HERE: adaptive fashion, dramaturgy and public making. Digital garment concepts, with fabrication and performance still in development.
- CRG-ANL: ongoing research on AI-native learning, instructional integrity and learner agency.
Who gets understood? Whose knowledge counts? Who keeps the authority?
coreyalejandro is my primary identity. coreyalejandroUX holds project work including the Bob hackathon kit. CRGSLab holds AI-native-learning and related research. These links connect the work; repository ownership and history remain in their current locations.
Toni Morrison: tell the ugly truth beautifully. Dave Chappelle: tell it with humor. Neil deGrasse Tyson: ask the hard questions about God, the universe and our place in it. Anna Wintour: fashion, business and their intersection. Andrej Karpathy: one of my muses. These are personal influences, not endorsements or affiliations.
I welcome remote work in AI Engineering, learning systems, instructional design, research engineering and human-centered AI evaluation.
| Path | Use This For | Link |
|---|---|---|
| Mothership | Governance system, registry, contracts, templates, truth state | The Living Constitution 2.0 |
| Public Site | Public-facing Living Constitution portfolio and Folio 001 | coreyalejandro.com |
| Module Status | What is working, partial, draft, planned, or unverified | MODULE_STATUS.md |
| Earlier Site Source | How the public site developed (the live site is built from newer source) | the-living-constitution-2.0-portfolio |
| Result | Repository |
|---|---|
| Pre-registered: a four-rule constitution cut Claude Haiku 4.5's judged rule violations from 22% to 0% (second judge: 27% to 2%); vague safety rules of the same length reached only 15% | constitutional-runtime-evals |
| Relational vs. transactional framing is readable inside Qwen2.5 models across vocabularies (81%, 85%, 92% for 0.5B, 1.5B, 7B; p = 0.001); steering it showed no causal effect, reported as such | eight-wonders-framing-probes |
| Relational Economy Benchmark: hidden-relationship worlds that reproduce byte for byte (159 tests); no language model scored yet | reb |
| PROACTIVE: a constitutional agent that blocks a flawed merge request (732 tests) | 28441830 |
| The Living Constitution 2.0 re-runs all four from fresh clones; 22 governance rules model-checked | the-living-constitution-2.0 |
I work at the intersection of:
- AI safety research
- runtime governance
- evidence systems
- human-AI interaction
- cognitive safety
- research-engineering infrastructure
- accessibility-aware tool design
- portfolio systems that distinguish verified work from drafts and prototypes
The Living Constitution 2.0 is the mothership of my current work.
It governs:
- C-RSP contracts
- module registries
- artifact registries
- route registries
- evidence records
- verification outputs
- research-to-paper-to-product templates
- public/private boundaries
- portfolio-safe public data
C-RSP means Constitutionally-Regulated Single Pass. It is one subsystem inside The Living Constitution 2.0, not the whole system.
| Surface | Status |
|---|---|
| The Living Constitution 2.0 repo | Working local governance/control-plane scope |
| Public site at coreyalejandro.com | Working static public surface |
| Folio 001 | Static prototype / origin exhibit |
| Research-to-paper-to-product template | Partial, usable template scaffold |
| HIDRS study repo | Active research module, partial |
I do not claim that every registered module is production-ready.
I do not claim that every research project is complete.
I do not claim that every public surface is final.
I do not expose private Lab mechanics, raw incident transcripts, unpublished drafts, secrets, or unreviewed evidence chains as public claims.
Research reviewers should start with The Living Constitution 2.0 and the system diagrams in its README.
Build reviewers should inspect the module registry, templates, verification scripts, and public site repo.
Evidence reviewers should start with the research results above, then MODULE_STATUS.md (a May 2026 snapshot), registry files, and public-safe artifacts.
Portfolio reviewers should start at coreyalejandro.com.
I am building a governed AI safety research-and-product system where claims, evidence, prototypes, working modules, and public-facing surfaces are separated by explicit truth status rather than collapsed into one polished but misleading portfolio.