Daniel Umana

What it is

ALIGN stands for AI Levels for Instructional Guidance and Navigation. It is a scale of six levels an instructor marks an assignment with, running from level 0, where no generative AI is used at any stage, up to level 5, where students design the AI workflow themselves. A student reads the level and knows what is allowed before starting the work.

Why we built it

I led the cross-functional faculty and staff team at Montgomery College that created ALIGN, and I co-authored the white paper, the faculty guides and the assignment-level checklists. Students were guessing from one course to the next about whether AI was allowed for brainstorming, for editing, for drafting, and whether any of it had to be cited. I think telling a student plainly what is expected of them is one of the kinder things an instructor can do, and the guessing is what costs them.

Who it is for

Faculty who want to state their expectations in a line instead of writing a policy of their own, and students who want to stop guessing. ALIGN is a communication tool. It does not require anyone to use AI, it does not rank courses, and it detects nothing. Open a level and read what it asks of the student.

What it is

A browser game where you take over AI policy for a fictional school district and run it for three school years, from 2026 to 2029. A campaign moves through four rounds and twelve decisions, and every decision shifts outcomes for seven groups of students, among them multilingual learners, students with IEPs and students with no reliable internet at home.

Why I built it

I built it for EDU 280, Social Justice and Urban Education, at American University, where Dr. William N. Thomas IV teaches the course. A policy can look fair on paper and still land hardest on the students with the least room to absorb it, and a seminar can agree with that sentence without anyone in the room having made a decision that produced it.

Who it is for

Instructors who teach policy, and anyone who has to make a call whose costs land unevenly. The dashboard opens on one district-wide number, and splitting that number by student group is a button you have to press, so the report at the end names the round where you first pressed it or tells you that you never did. Take office and play a round.

What it is

A KWL chart is a classroom staple. Before reading, a student writes what they know about a topic and what they want to know, and afterward they write what they learned. reKWL is that chart with a language model behind it, and the model never fills a square in for the student. It reads what was written and comes back with questions.

Why I built it

Most reading tools answer the student, which hands over the part of the work that was supposed to be theirs. reKWL asks instead, and it tunes the questions to one of four reading levels, from grades one through three up to bachelor’s and doctoral work, so the same passage comes back differently for a fourth grader and a doctoral student. The research behind that progression is named on the app’s own page.

Who it is for

Teachers and instructors from elementary school through graduate school who assign reading and want students working on it rather than collecting an answer. Pick a level and run the sample passage, then switch the level and read the questions again.

What it is

The site that carries a sixty minute workshop I ran on using AI for everyday work, the kind of work sitting on someone’s desk anyway: writing more clearly, organizing a task, comparing options, checking an assumption. Its sections hold the session’s main rule, its guardrails and its practice prompts, and a guide built out of what a participant typed downloads at the end.

Why I built it

A workshop ends and the handout goes in a drawer. I wanted the prompts and the guardrails to sit somewhere participants could reach the week after, when they finally hit the task they wanted the help with. The site also puts the Montgomery College technology policies into plain language, since what people ask first is what they are allowed to type into the tool.

Who it is for

Working adults using AI on the job rather than in a classroom, and anyone who runs a session like this and wants the materials to outlast the hour. Start at The Main Rule, then open Guardrails.

What it is

The companion site for a sixty minute session Dr. Angela Lanier and I built for Montgomery College faculty during Professional Week, on using AI while designing an assignment. It holds the premise of the session, the college policies in plain language, two worked demonstrations and a toolkit of prompt cards a professor can copy straight into their own course.

Why I built it

Faculty were being told to use AI in assignment design without being told what to do with what it gives back. The premise we teach is that the instructor supplies the expertise and the model supplies material to judge, so a card asks it to test an assignment for clarity, or to simulate how a student might read the prompt, or to check the wording against the rubric. No card asks it to write the assignment.

Who it is for

College faculty writing or revising assignments, and the staff who support them. Read the premise, then open the Toolkit, which is the part faculty take back to their courses.

What it is

DoP is short for Dissertation of Practice, the applied dissertation my doctoral program is built around, and this is the self-assessment that goes with it. Students rate themselves against the nine competencies and 33 sub-elements the program expects them to develop. The app scores that as they go, names the priority competency in each category, writes reflection prompts out of the answers they gave, and generates two PDFs, one a filled copy of the original artifact and one a results summary.

Why I built it

The instrument was a paper form. A student filled it in, added up a score and got back nothing that pointed anywhere. I rebuilt it for my own cohort in EDU 703 at American University so the same questions return something usable in the moment, and the arithmetic is plain and deterministic, with no model called at any point and nothing about an answer leaving the browser.

Who it is for

Doctoral students taking honest stock of where they stand, and any program still running this kind of inventory on paper. Score yourself and read what comes back. There is no login and no account.