I design joyful, standards-aligned learning for children who can see it before they can name it.

The problem

Abstract calendars do comparatively little work for children of this age. “Yesterday” can mean last week. What does work is the body and the senses — the coat that goes on before outdoor time, the dark that arrives before pickup. The design problem was to take a real astronomical event and make it something a four-year-old could see, feel, name, and belong to.

My role

I designed and taught this unit in a prekindergarten classroom, documented it, and then developed it into the guide. I authored all seven lesson plans, the standards alignment, the slide deck, and every appendix. All illustrations, diagrams, templates and page ornaments were drawn originally for this edition — no stock art. Four of the six songs are mine; the other two are traditional.

Design decisions

1. Science before culture. The flashlight-and-ball investigation sits in Week One rather than later, because children need the physical experience of the slanting beam before the cultural material will mean much to them. Week Two’s festivals then answer a question the science has already raised.
2. The art project has a destination. Every votive a child makes in Lesson 4 is carried into the Spiral Ceremony in Lesson 7 — and children know that from the start. It turns an open-ended art experience into a genuine materials investigation with a reason.
3. Rehearse the shape before the event. The Spiral Center opens in Week Two, not Week Three, so that by the time the ceremony arrives the spiral is already familiar and the room stays calm.
4. Culture as culture, not observance. Dongzhi, Yalda, Yule, Saturnalia, St. Lucia Day and Inti Raymi are presented as cultural information. The unit does not teach religious practice, and family participation is genuinely optional.
5. Assessment named as observable behavior. Each experience is mapped to Work Sampling, Teaching Strategies GOLD and HighScope COR — but the guide says plainly that programs on another system should substitute their own indicators, because the observable behavior is the part that matters.

Evidence & iteration

Everything in the guide has been used with four-year-olds. A daily sunset chart begins on day one and runs the full three weeks, so the children collect the evidence themselves and carry it home on Friday. This is the second edition: the ceremony now specifies battery lights so no family arrives worried about candles, and Appendix F carries a print-at-100% warning because the cut-out templates come out the wrong size otherwise.

The problem

On a screen, art appreciation collapses easily into watching. And in a Spanish-language lesson there is a second risk: a child who is still building language will say nothing at all if the first thing asked of them is a full sentence about a painting. The lesson had to give children something to notice, something to say, and something to make — in that order.

My role

I wrote and designed the whole lesson in Spanish — the artist selection, the works shown, the observation protocol, the studio task, the self-assessment checklist, the critique language, and the closing artist statement frames.

Design decisions

1. Silence comes first. Thirty seconds of mirada silenciosa before anyone speaks, then four sentence stems — Veo… Me pregunto… Siento… Me recuerda a… — so every child has a way into the language, not just the confident ones.
2. Four lenses, reused three times. Color, repetición, material, sentido are introduced as observation questions, then reappear as planning questions and again as critique questions. One vocabulary set carries the entire lesson, which is what makes new terms stick.
3. “No copies una obra.” The studio task takes an idea, not an image. That single instruction turns a craft activity into a design problem, and it is what the artist statement is later assessed against.
4. An artist chosen for her materials. Falcone works with spice pigments and latex balloons as well as paint — so “what is it made of?” is a real question with a surprising answer, and the cultural content is contemporary and specific rather than generic.
5. Assessment that is also language production. The lesson closes on frames: Mi obra expresa… Elegí estos colores porque… Repetí esta forma para… The evidence of learning and the Spanish output are the same artifact.

Evidence & iteration

Children self-check against a five-item lista de control del artista before the gallery close — including “I can explain at least one artistic decision.” The peer feedback protocol requires an observation before any suggestion (Noto… before Me pregunto si…), which is what keeps a five-minute critique from turning into a list of fixes.

The problem

Spanish orthography is shallow: roughly twenty-four phonemes map to their graphemes with near-perfect regularity, and the open CV syllable dominates. A child who knows five vowels and two consonants can read real words in the first weeks. That generosity is a trap for authors — it makes it easy to write text that is technically decodable and instructionally flat, or text that quietly admits sounds nobody has taught.

My role

Sole author. Four sequenced 10–15 minute lesson plans with teacher script in English and Spanish, differentiation and formative assessment criteria; one student-facing passage with a complete word-level decodability audit; print-ready syllable cards, word cards and a sentence strip; and framework notes covering orthography, syllable taxonomy, accentuation, a 24-point editorial QA checklist, and a bilingual glossary for editorial correspondence.

Design decisions

1. Write to the cumulative inventory, not to the lesson. The legal set at any point is every element taught to date plus the taught high-frequency list. Writing to the current lesson alone produces a page of ma me mi. Writing to the full inventory is what makes connected, meaningful text possible this early — the difference between a worksheet and a book.
2. Stop at four lessons on purpose. It is far easier to demonstrate coverage of a large inventory than control over a small one, and control is what decodable authoring actually asks for.
3. Guard the schwa. /m/ is a continuant, so it can be stretched — which is exactly why teachers add one: “muh” instead of /mmmm/. That extra vowel is invisible now and is the reason blending breaks down three lessons later, so the script says so out loud.
4. Keep /r/ out. Spanish carries two rhotics, and word-initial r is always the trill. Rosa and mira look harmless beside mesa, but each smuggles in a phoneme, a grapheme and a positional rule that has not been taught. This is the single most common error in early Spanish decodables.
5. Audit at the syllable, and publish the audit. The Lesson Four passage ships with a word-by-word table — syllables, structure, elements, status, repetition count — so an editor can check the claim instead of taking it.

Editorial QA

Twenty-four checks in four bands run against a manuscript before it leaves the desk: sequence compliance, decodability metrics, language quality, and comprehension and art. The language band is the one machines cannot run — does it read as Spanish rather than as translated English, are subject pronouns over-explicit, does the register carry adultisms in dialogue, does the illustration support meaning without letting a child identify the word from the picture.

My role

I evaluate AI-generated scoring against a structured, research-based rubric, using SEL and education expertise to judge the quality of authentic learner responses. Alongside the review work I contribute Spanish localization, captioning, video editing, dataset preparation, and product-quality feedback.

How the work moves

1. Study the framework. Understand what the rubric intends and where its distinctions actually fall.
2. Review learner evidence in context. Read the open-ended response as a whole, not as a string of keywords.
3. Apply expert judgment. Distinguish depth, signal and sophistication.
4. Compare human and AI. Agree or disagree with the automated score, explicitly.
5. Identify patterns and edge cases. Notice ambiguity, prompt problems, recurring misses.
6. Feed it back. Document findings so they can change the product, not just the score.

Where human judgment matters

Developmental context. Recognizing a meaningful idea when the articulation is still emerging — a five-year-old’s real insight rarely arrives in complete sentences.

Depth versus surface. Separating authentic understanding from language that only appears to fit the rubric.

Ambiguity and calibration. Identifying the places where two reasonable reviewers would legitimately diverge, and saying so rather than smoothing it over.

What data changed

Across repeated reviews I look for patterns in the model’s scoring and in how learners interact with the content itself. When a prompt repeatedly fails to elicit the intended response, or when a scoring distinction produces recurring ambiguity, that is a content problem wearing a model problem’s clothing — and it gets flagged and documented for the product team as such. In one project I reconciled a messy multi-source dataset with inconsistent inputs and poor translations, manually aligning it so it could support downstream product work.

Bilingual product quality

Spanish translation and localization, captioning, video editing, dataset preparation and content-quality decisions. I weigh register, dialect, age appropriateness, terminology and naturalness — especially in short-form learning experiences, where small language choices carry a lot. The principle: accuracy is necessary; naturalness is equally important.

The problem

Most feelings cards ask a child to name an emotion and then stop. That produces guessing: the child hunts for the answer the adult wants. The harder and more useful task is inference from evidence — reading a face, a body, a situation — and that requires stories with enough context to argue about.

My role

I wrote both editions, designed the layout, and built the standards crosswalk. Fourteen illustrated situations, a teacher guide, and an alignment reference, in English and in Spanish.

Design decisions

1. Open-ended on purpose. There may be more than one reasonable answer. A child may name several feelings, change their thinking mid-conversation, or connect the scene to something that happened to them — and the guide states that all three are meaningful signs of development, not errors.
2. Ask for evidence, not for the answer. The two anchor prompts are “What do you notice?” and “What makes you think that?” That single move turns a feelings card into an inference task with oral-language work attached.
3. Localize the children, not only the words. In the Spanish edition the characters are renamed — Evie becomes Emilia — so a Spanish-speaking child meets children with names from their own world. Same illustration, same situation, different reader.
4. Include the hard feelings. Exclusion (“You can’t play”), a lost lovey, a friend who keeps going after “stop,” a grownup who leaves for the afternoon. Contentment and delight are in there too, but a set that only covers happy is not usable on the day a child needs it.
5. Alignment as a guide, not a claim. The crosswalk is explicit that it is an alignment guide, and not evidence that a child has mastered a standard after using a single story.

Standards crosswalk

Mapped to Head Start ELOF (P-SE 6, P-SE 7), Common Core Speaking & Listening for Kindergarten (SL.K.1, SL.K.2, SL.K.3, SL.K.6), and the California Preschool/Transitional Kindergarten Learning Foundations (2024) — SED 1.3 understanding emotions, 1.4 regulating emotions, 1.7 understanding others’ thoughts and experiences, 1.8 empathy and caring.

The problem

A visual schedule earns its place when a child can read it across the room and use it without an adult. The original set fought itself on that measure: a zigzag border and decorative frame competed with the illustration, the library was too small to cover a real classroom day, and “editable” meant a file you could open rather than a file you could actually change.

My role

Designer and author of both editions — illustration, layout, the card taxonomy, the print system, and the seller-facing preview.

Design decisions

1. Keep the concept, replace the craft. Same font, cleaner border, sharper artwork. Teachers who already owned the set had to still recognize it — a redesign that reads as a different product loses the people it was for.
2. One picture, one word. Every card pairs a single clear illustration with a short label on a white ground inside one calm frame. Nothing extra to decode, because the children who lean on a visual schedule hardest are the ones least able to filter a busy border.
3. Children of every background, throughout. Skin tones, hair, families and mobility aids are represented right through the set, not in a diversity page at the back.
4. Editable means actually editable. Label, border and illustration are separate selectable objects, plus blank templates in both sizes — rename, replace the art, restyle the frame, duplicate the page — in PowerPoint, Keynote or Google Slides.
5. Organize by classroom flow, not alphabet. Seventy-two routines across eighteen pages grouped as Start + Care, Learning, People + Community, and Creative + Play, so related cards print together and a teacher can prep one part of the day in one pass.
6. Design for the laminator. Four cards to a page, true size at 100%, trimmed on faint dashed guides, fitting a 4 × 6 pouch. The same order in the PDF and the PowerPoint, so the two never drift.

Before → after