By Sarthak Sharma — Full Stack Developer, Kathmandu, Nepal
A few months ago I wrote about how Kung Fu Quiz and HomeschoolAsia are revolutionizing EdTech in Asia. That post was, admittedly, a developer's post — it was about architecture, WebSockets, the YouTube IFrame API, and the satisfaction of shipping features that scale. I stand by every word of it.
But in the months since, readers, teachers, and a few students have asked me a very different question. Not "how did you build it?" but "does it actually help?" And that is a far more interesting question, because it forces me to take off the developer hat and put on the one I care about more: the hat of someone who genuinely wants learners to do better.
So this is a different kind of post. This is not about the code behind HomeschoolAsia, Kung Fu Quiz, or CIE Past Papers. This is about the learning. I've had the privilege of contributing to the development of HomeschoolAsia and Kung Fu Quiz, and I maintain and update CIE Past Papers using a Laravel CMS. That vantage point — sitting between the product and the people who use it — has changed how I think about what "good" educational technology even means. I want to share what I've seen.
The Uncomfortable Truth About Most EdTech
Let me start with something that took me years to fully accept: technology does not make people learn. If anything, badly-designed technology makes learning worse by manufacturing the feeling of progress without the substance of it.
We've all done it. You watch a two-hour lecture on YouTube, nod along, feel productive, and then discover a day later that you can't reproduce a single idea from it. That's not a character flaw. It's how memory works. Cognitive science has been extremely clear on this point for decades. In one of the most cited studies in the field — Roediger and Karpicke's 2006 "Test-Enhanced Learning" experiments (Psychological Science, Vol. 17, No. 3) — students who repeatedly took recall tests on a prose passage retained 61% of the material a week later, versus 40% for students who simply restudied it. Strikingly, the repeated restudy actually made students more confident in their ability to remember — even as they remembered less.
That single finding — that retrieving information beats reviewing information — is, in my opinion, the most important idea in all of education technology. And most platforms ignore it completely. They optimize for content delivery: more videos, more PDFs, more hours of footage. But delivery is not learning. Retrieval is learning. Feedback is learning. Struggle, followed by correction, is learning.
The three platforms I want to talk about each attack a different failure point in that chain. And once you see them through that lens, they stop looking like three unrelated products and start looking like three parts of one honest answer to the question: how do we get students to actually learn, not just consume?
Kung Fu Quiz: Making It Impossible to Watch Passively
Kung Fu Quiz begins with a simple, almost stubborn refusal to let students watch a video passively.
YouTube is the largest free educational library humanity has ever assembled. People watch over a billion hours of video a day on it, and a huge share of that is people trying to learn something. Teachers know this. Students live on it. But the platform was never built for learning, and it shows: attention drops off a cliff after a few minutes, the sidebar is engineered to distract you, and there is no mechanism whatsoever to check whether anything is sticking.
Kung Fu Quiz takes any YouTube video and turns it into an interactive challenge. A teacher — called a "Sifu" in the platform's martial-arts framing — embeds questions directly into the video timeline. The video pauses at a chosen moment, the student has to answer, and only then does playback resume. Students join a session using a "Dojo Code," and depending on the mode the teacher picks, they either compete on a live leaderboard, get immediate corrective feedback, or work through a self-paced assessment.
Here's why this matters, and it's not marketing — it's research. When you insert questions into a video lecture, students report far fewer episodes of mind-wandering than when they watch the same video uninterrupted. And crucially, the benefit isn't limited to the questioned material. Studies of interpolated testing find that students in the tested condition pay more attention to the content between the questions too — a forward-reaching effect that lifts learning across the whole video. Interactive videos also see meaningfully higher completion rates than their passive equivalents, because each question creates a small commitment that pulls the learner forward.
From where I sit, the design decisions that make this work are subtle. The platform offers six distinct challenge formats — single-answer and multiple-answer questions, drag-words-into-a-box, fill-in-the-blanks, sort-the-sentence, and a "button" type that can redirect students to a key section or resource. That variety isn't decoration; different question types exercise different kinds of recall, and mixing them keeps the cognitive work honest. There are also multiple "Dojo Modes" — Game, Assessment, Feedback, and a Live mode for hosting a quiz on a big screen — so a single quiz can be repurposed for a high-energy classroom competition one day and a quiet diagnostic check the next.
More recently the platform added an AI generator called sensAI that takes a YouTube link and auto-generates timestamped, video-aligned questions in minutes, including in languages beyond English. When I think about a single teacher in a busy school with no prep time, that feature is the difference between "I'll do this someday" and "I'll do this tonight." The friction of quiz creation was always the thing standing between good intentions and actual interactive lessons.
The martial-arts theme — Sifus, Dojos, leveling up — sounds like a gimmick until you watch students respond to it. Gamified learning consistently improves engagement and retention when it's done well. A 2023 meta-analysis (Huang, Maldonado et al., published in Frontiers in Psychology, PMC10591086) pooled 41 studies with 49 independent samples and more than 5,071 participants and found "an overall significant large effect size (g = 0.822 [0.567 to 1.078])" for gamification on learning outcomes. The key phrase is "done well" — the same analysis found the effect was strongest for interventions that ran longer than a semester. Points and leaderboards bolted onto boring content don't help. But a coherent narrative that reframes a quiz as a challenge rather than a test changes the emotional texture of the whole experience, and that emotional texture is what determines whether a student comes back tomorrow.
For educators, the quieter feature is the data. Kung Fu Quiz gives teachers individual and group insights after each Dojo, so they can spot exactly where a class collectively stumbled. That's formative assessment in the truest sense — assessment that happens during learning so a teacher can adjust while it still matters, rather than discovering the problem on the final exam when it's too late to fix.
HomeschoolAsia: Ending the Scavenger Hunt
If Kung Fu Quiz solves the passivity problem, HomeschoolAsia solves a problem I think of as the "scavenger hunt."
Any student who has prepared for Cambridge IGCSE or A Levels knows this pain intimately. The syllabus lives on one site. The revision notes live on a forum somebody linked in 2019. The past papers are scattered across three different repositories with broken links. Video explanations are on YouTube, if you can find a good one. Mock exams — well, you mostly just hope. The single scarcest resource for an exam student isn't information; it's time, and the scavenger hunt burns it mercilessly.
HomeschoolAsia consolidates the entire Cambridge journey into one exam-focused platform. It covers the core IGCSE subjects — Biology (0610), Chemistry (0620), Physics (0625), Additional Maths (0606), and Extended Maths (0580) — and the major A Level subjects including Biology (9700), Chemistry (9701), Physics (9702), Mathematics (9709), Accounting (9706), and Economics (9708). For each, students get a coherent stack of resources instead of a pile of disconnected links.
The centerpiece is the Learning Hub, built around a library of 1,500+ animated interactive videos mapped directly to the Cambridge syllabus. This is where the platform's DNA shows: the videos aren't passive lectures but quiz-embedded lessons, carrying the same active-recall philosophy that powers Kung Fu Quiz. Around them sit the pieces that make self-study actually work:
- Downloadable revision notes for every topic, the kind students reach for right before a test to lock in definitions and formulas.
- Mock exams with diagnostic reports. This is the feature I'd point to first. Diagnostic assessment — figuring out precisely what a student does and doesn't know — is one of the most evidence-backed tools in education, and it's exactly what self-directed learners lack. A student studying alone has no teacher to say "you're weak on electrolysis, focus there." The diagnostic report is that teacher.
- Topical past papers and a "Generate Testpaper" tool, so students can drill a specific weak topic instead of grinding whole random papers.
- Tutor Support, where subject experts give exam-style evaluations and answer queries — a human in the loop, which matters more than any feature list suggests.
- A built-in Planbook and Learning Guides with study timetables, because the hardest part of self-study is often just knowing what to do next.
There's a business-model detail here that I think is genuinely important for educational equity: HomeschoolAsia has a meaningful free tier. Students get free tutor credits on signup and free access to a real slice of the resources before deciding whether to pay for the premium plan (priced at SGD 49.75/month, or SGD 16.58 per subject per month on the custom plan). In a region where private tutoring is expensive and unevenly distributed, the difference between "free" and "paywalled" is the difference between a student getting help and not. The cost of one-on-one tutoring is precisely the barrier that keeps personalized support out of reach for most families; anything that lowers it expands who gets to succeed.
There's a wider context that makes all of this land harder. Demand for Cambridge International qualifications is surging across South and Southeast Asia. Per Cambridge International Education's own August 2025 announcement (via PR Newswire), "more than 165,000 entries were made for Cambridge qualifications this June [2025]" in the Southeast Asia and Pacific region — a nearly 4% year-on-year rise, part of over 680,000 students worldwide receiving June 2025 results (a 9% increase). And according to the 2024 Cambridge Student Destinations survey, "globally, 89% of the 2024 Cambridge International AS & A Level student cohort progressed directly to university after graduating from school, 47% in their own country and 42% overseas." For a Nepali or Southeast Asian family, a strong set of Cambridge results is a genuine passport to global opportunity. The stakes are enormous — and the support infrastructure has historically been thin. That gap is exactly what HomeschoolAsia was built to fill.
CIE Past Papers: The Discipline of Real Practice
The third platform is the one I'm closest to on a day-to-day basis, because I maintain and update CIE Past Papers using a Laravel CMS. It is also, deceptively, the simplest — and I've come to believe simplicity is its greatest strength.
CIE Past Papers is a free, well-organized repository of Cambridge past papers for IGCSE and AS/A Levels, running from 2020 up to the latest series. But it's not just question papers. For each subject and session it pulls together the marking scheme, the examiner report, solved past papers with written solutions and video walkthroughs, and the exam timetable. You can read everything in a built-in PDF viewer or download it for offline study — which matters a great deal in places where internet access is intermittent. You navigate by subject, year, and session, and you're practicing within a few clicks.
Why does this deserve its own platform? Because past-paper practice is, quite simply, the highest-leverage exam-prep activity that exists, and most students do it badly. Working through real papers under timed conditions familiarizes students with the exam format, sharpens time management, and surfaces knowledge gaps while there's still time to close them. Familiarity itself reduces exam anxiety: as the British Psychological Society notes, repeated exam practice with past papers reduces the effects of anxiety on cognitive resources "through rehearsal and increasing familiarity." When the format holds no surprises, a student can spend their nerves on the questions instead of the unknown.
But here's the part most students miss, and the part the platform quietly fixes: the marking scheme is the most valuable document in the entire pile, and almost nobody uses it properly. Students download it, glance at it once as an answer key, and move on. That's a waste. A marking scheme isn't a model essay — it's a map of how examiners award marks. It shows you that in Chemistry, "oxidised" earns the mark where "gained oxygen" does not; that keywords beat waffle; that structure earns credit even when the final number is slightly off. Learning to read the mark scheme like an examiner is one of the fastest ways I know to move a student from a B to an A without teaching them a single new fact.
The examiner reports are the second underused goldmine. They document the exact mistakes real students made on real papers — the misread command words, the skipped units, the misconceptions. By having marking schemes, examiner reports, and solved papers sitting next to the question papers, CIE Past Papers turns a passive download into an active feedback loop. And that feedback loop is, once again, the same principle underneath everything: retrieve, check against the standard, correct, repeat.
Maintaining this site has taught me a humbling lesson about EdTech. The flashiest feature is rarely the most valuable one. A reliable, fast, well-categorized, always-up-to-date past-paper library with the right supporting documents attached is worth more to a stressed student in April than almost any amount of AI wizardry. My job on CIE Past Papers is mostly invisible: keep it current, keep it fast, keep it free, keep it correct. That invisibility is the point.
How the Three Fit Together
What I didn't fully appreciate until I'd worked across all three is that they map almost perfectly onto the three phases of how a student actually learns a subject.
- Encounter and engage — This is Kung Fu Quiz territory (and the interactive videos in HomeschoolAsia's Learning Hub). You meet new material, and instead of letting it wash over you, you're forced to engage with it actively, question by question. First contact becomes first practice.
- Consolidate and structure — This is the heart of HomeschoolAsia. Revision notes organize what you've met, diagnostic mock exams tell you where you stand, tutor support unblocks you when you're stuck, and the planbook keeps you moving through the syllabus in order. Scattered exposure becomes structured mastery.
- Rehearse and perform — This is where CIE Past Papers comes in. You take everything you've built and rehearse it under real exam conditions, then use the marking schemes and examiner reports to close the final gaps between what you know and what you can demonstrate on paper.
The connections aren't just conceptual, either. The platforms are genuinely cross-linked: HomeschoolAsia points students to CIE Past Papers for practice, and CIE Past Papers links back to HomeschoolAsia's Learning Hub for the videos, tutor support, and testpaper tools. A student can move through the full arc — encounter, consolidate, rehearse — without ever falling into the scavenger-hunt trap.
This is also, I think, the answer to the passivity problem at a system level. Any one of these tools used in isolation could still be used passively. But used together, in sequence, they form a loop that keeps pulling the student back toward active retrieval and honest feedback. That loop is the whole game.
What This Looks Like for Real Learners
Let me make this concrete, because frameworks are easy and students are hard.
Picture a Grade 11 student in Kathmandu — or Yangon, or Dhaka — preparing for A Level Chemistry with no access to an expensive tutor. Six months out, she works through the interactive video lessons, and because the questions are embedded, she can't fool herself into thinking she understands something she doesn't. Three months out, she runs the diagnostic mock exams on HomeschoolAsia and discovers her organic chemistry is far weaker than she thought — a blind spot she'd never have found on her own. She spends her tutor credits getting exam-style feedback on exactly that topic, drills it with topical testpapers, and then, in the final weeks, moves to CIE Past Papers to sit full timed papers. She marks her own work against the scheme, reads the examiner report, and realizes she keeps losing marks not on knowledge but on command words — "explain" versus "describe." She fixes it. That's not a hypothetical; that's the exact path these tools are designed to enable, and it's a path that used to require money and connections she may not have.
Now picture the teacher on the other side. A Sifu with 40 students and no time creates a Kung Fu Quiz from a YouTube video in a few minutes with sensAI, runs it live on the projector, watches the leaderboard light up, and gets a report afterward showing that three-quarters of the class missed the same concept. Tomorrow's lesson writes itself. That teacher didn't get replaced by technology; they got a superpower.
What Building and Maintaining These Taught Me
A few honest reflections, from the inside.
Free tiers change who gets educated. I don't say this as a marketing line. Across South Asia, the binding constraint on educational opportunity is very often cost. Every meaningful thing you can put behind "free" — a past paper, a diagnostic report, a starter set of tutor credits — expands the set of students who get a fair shot. That's a design decision with a moral dimension, and I'm proud the platforms take it seriously.
Domain expertise beats cleverness. You cannot build good Cambridge preparation tools by modeling the data structures of the syllabus. You have to understand the curriculum, how it's assessed, and how students fail. The most valuable features — diagnostic reports, mark-scheme-adjacent practice, command-word awareness — come from understanding the pedagogy, not the technology.
The boring work is the important work. On CIE Past Papers, keeping the library current, fast, and correct matters more than any shiny addition. Reliability is a feature. For a student the night before an exam, a site that loads instantly and has the right paper is worth more than a dozen demos.
Engagement and rigor are not opposites. The martial-arts mascots and leaderboards aren't in tension with serious learning — they're in service of it. Delight is load-bearing. If a student doesn't come back, the best pedagogy in the world does nothing.
A Note on Context — and Honesty
I want to be careful not to oversell. The reach and impact figures that circulate in this ecosystem — student counts, country counts, distinction rates like "over 70% of students scored distinctions" — are largely self-reported and vary across sources, so I treat them as directional rather than gospel. The research on gamification is genuinely positive but also mixed, with effects that depend heavily on how long the intervention runs and how well it's designed. And none of these tools remove the two real constraints that shape EdTech across Nepal and South Asia: uneven internet access and the deep urban-rural divide. A brilliant platform is useless to a student without a stable connection or a device. Internet penetration in Nepal climbed from under 1% in 2005 to roughly two-thirds of the population by 2021, which is why platforms like these are viable at all — but the last mile is still very real, and honesty about it matters more than optimism.
What I can say with confidence is that the design philosophy is right. These platforms bet on active recall over passive consumption, on feedback over content dumps, on structure over chaos, and on access over exclusivity. Those are the right bets. The learning science backs every one of them.
Closing Thoughts
When I wrote my earlier post, I was proud of the engineering. I still am. But the longer I work in this space, the more I believe the engineering is the easy part. The hard part is respecting how humans actually learn — and refusing to ship the comfortable illusion of learning in its place.
Kung Fu Quiz refuses to let students watch passively. HomeschoolAsia refuses to make them hunt for scattered resources or study blind. CIE Past Papers refuses to let practice be anything less than real. Each one, in its own way, closes a gap between feeling like you're learning and actually learning.
If you're a teacher, try turning your next YouTube lesson into a challenge on Kung Fu Quiz and watch what happens to the room. If you're an IGCSE or A Level student — or the parent of one — start with the free resources on HomeschoolAsia and the past-paper library on CIE Past Papers, and build the encounter-consolidate-rehearse loop for yourself. It costs nothing to begin, and it works.
That, more than any line of code I've written, is what keeps me in this field.
— Sarthak Sharma