Learning Experience Designer, 10+ years in EdTech. I built the curricula at Yandex Practicum, TripleTen and BEAM Institute, and then the tooling those content teams ran on: 4C/ID pipelines, internal automations, an LMS shipped solo.
Learning Experience Designer
Two tracks came off one curriculum, a full career change and a shorter one for designers already working. A pipeline of designers, developers and reviewers put out a cohort every four to six weeks, with outcomes written against what the industry was hiring for.
Learning Experience Designer
Every unit ended in something a switcher could show: brand identity, editorial layout, packaging, social. Weekly critiques and SME reviews led to a portfolio defence at the end of the track.
Learning Experience Designer
Learners worked real briefs from partner companies across performance, content, lifecycle and analytics. Assessments mirrored the tasks a junior marketer gets, and practice was scored against funnel numbers.
Learning Experience Designer
Each client got a short course recut from TripleTen's master management curriculum to fit their roles and their business. Some were localised for LatAm, and the video segments ran on HeyGen AI-avatar scripts.
Learning Experience Designer
Each client's course started from TripleTen's long-form backend program and was cut to their stack and their seniority mix. Where a real codebase was available, assessments were built on it.
Learning Experience Designer
Prompt engineering, the AI workflows for a given role, and where the guardrails sit. Each version was tuned to the tools a client already ran and the policies they had to honour.
Learning Experience Designer · Freelance · project-based
Nothing existed before this one: needs analysis, curriculum architecture, content and QA all ran from zero. Consultants from Billigence, a data-governance firm, kept the regulatory material current.
Learning Experience Designer · Freelance · project-based
The track was written with practising PMs, so it taught where AI belongs in discovery, planning and delivery. Practice ran on their own artefacts: briefs, roadmaps, retros, each rebuilt around an AI workflow.
AI Consulting · L&D · Internal Tooling · Freelance · current
Workshops take the instructional-design team of a K-3 maths platform through AI in content design. Mentoring covers prompt engineering and review pipelines one to one. The tooling comes with it: slide and board generators, autofill userscripts, a lesson-review pipeline, an adaptive-curriculum knowledge graph.
Recent work, selected. 10+ years, 50+ courses, more clients than fit on a page.
A new SDR's desk, running in the browser: Outreach, Gong, Salesforce, LinkedIn, Calendar, a dialler and Slack, each one a draggable window. Have a look around.
Open the desktop →
Three SCORM 1.2 packages, one brief each: security onboarding for new hires, de-escalation for a call centre, an ERP procurement rollout. Each runs from the analysis brief to a package you can drop into an LMS.
See all three →
A payments risk desk: flagged transactions, a risk score, the rules that fired, and Approve, Decline or Escalate. Every shift is generated fresh, and the model is wrong often enough that you have to read the signals yourself.
Open the queue →
An attacker phones posing as IT Security. He is voiced, your replies branch the call, and his pressure adapts: urgency, authority, threat. Four endings.
Take the call →
A card-testing attack is running and you are the only analyst on call. Chase the IPs and the bot rotates. Six endings across three live meters: fraud losses, customer impact, leadership trust.
Run the night →
Security and privacy awareness, the usual 40-minute deck rebuilt as six self-contained micro-units on a retrieval chain. The case page is the design map; Unit 1 is playable in full, with a 2-minute film, a longread, nine checks and an inbox simulator.
▶ Play Unit 1 →
Fly round the launch field in 3D and pick a building. Each one is held until the job inside it is done, and all four ask for something different: bin the virus files without binning ours, reconnect four wires, hold a purge down while it fights back, pull a lever and keep it there.
Take the field back →
Click, double-click, drag, chase. The four movements a six-year-old needs before any of the rest of a computer makes sense, each at three speeds, scored out of three stars. Dressed as a game because a drill dressed as a worksheet gets opened once.
Play a round →
Hang a photo in a room and a machine says what it thinks it sees, then decides whether it belongs. The recogniser is MobileNet running inside the page, so there is no account to make and the photo never leaves the computer. It shows its guesses and the score it judged on, so a child can see why it said no.
Hang a photo →Twenty of these live in one place: a Windows file manager, a Chrome tab strip, a Google Docs find-and-replace, a pixel-art canvas. Open the catalogue →
A place to run the whole business of teaching, instead of stitching together a video host, a quiz tool and a marking spreadsheet.
A Flask backend and a set of Tampermonkey scripts fill matching, categorisation and fill-in-the-blank tests on the teacher-training platforms. Each instructional designer gets back about five hours a week that used to go into clicking.
The bot pushes homework prompts into Telegram, collects what comes back and lays it out for instructors in a triage view. Learners stay in the app they already have open.
The add-on carries the branded template, a catalogue of layouts and a custom maths font into Google Slides. A deck that took two hours now takes about fifteen minutes, and it still comes out on brand.
Hand it a lesson-plan JSON and it returns a Google Slides deck already on the branded template. The layout step disappears, so a designer opens the file and goes straight to the content.
Structured input goes in and the tool draws the board in Miro or Figma over the API, whether that is a workshop canvas, a lesson canvas or a retro. Brand config and layouts live inside it, so two teams asking for the same board get the same board.
A knowledge graph holds the skills of K-3 maths and the observable objectives sitting under each one. It feeds the diagnostic, the BKT mastery estimate, and the lesson a given learner gets next.
The dashboard catalogues lesson assets, tracks what has been reviewed and keeps the methodology references where the curriculum team can reach them. MCP endpoints expose the same data, so AI tools read from what the team reads from.
Slack threads, Notion comments, mentor rubrics and post-cohort survey CSVs all land in one place, and recorded lessons are transcribed alongside them. The pipeline normalises the lot into an issue database the QA team can work through.