Technology
Adaptive Learning for Trade Certification
VoltExam isn't a static question bank. We're building an intelligent learning engine that adapts to each user's knowledge gaps, optimizes study time, and predicts exam readiness — all running on-device so it works on job sites with no connectivity.
The Problem
Trade certification exams cover hundreds of topics, but every learner has different strengths. A journeyman electrician studying for their master license already knows Ohm's Law — they need to focus on NEC code changes and load calculations. Traditional apps treat every user the same, wasting study time on material they've already mastered.
Our users are working professionals who study in 10-minute windows between jobs. They can't afford to waste a single session on questions they already know. Every minute of study time has to count.
Adaptive Learning Engine
At the core of VoltExam is a performance modeling system that builds a real-time knowledge map for each user across every topic domain in their trade. The engine continuously updates difficulty estimates and user proficiency scores to surface the questions that will produce the largest learning gains per minute of study.
Knowledge state tracking — Every answer updates a per-topic proficiency model. We track not just whether you got it right, but response time, confidence patterns, and decay curves to distinguish genuine mastery from lucky guesses.
Difficulty calibration — Question difficulty isn't static. We use response data across our user base to continuously calibrate difficulty ratings using Item Response Theory (IRT) models, ensuring each question is appropriately challenging for the learner encountering it.
Optimal question selection — Given a user's current knowledge state and available study time, the engine selects questions at the frontier of their understanding — the zone where learning is fastest. This is inspired by research on the "zone of proximal development" adapted for adult professional learners.
Exam readiness prediction — Our Am I Ready? feature uses the knowledge state model to estimate the probability of passing the actual certification exam, giving users a data-driven signal for when they're prepared.
Spaced Repetition & Memory Optimization
VoltExam implements a modified spaced repetition algorithm tuned for professional certification material. Unlike generic SRS systems designed for language learning, our intervals account for the structured dependency graphs that exist in trade knowledge — you can't understand transformer sizing without understanding voltage drop first.
Dependency-aware scheduling — The system understands prerequisite relationships between topics. If you forget a foundational concept, dependent topics are automatically resurfaced before their scheduled review date.
Forgetting curve modeling — We model memory decay per user per topic, using actual response data rather than fixed intervals. Users who demonstrate strong initial encoding get longer review intervals automatically.
Offline-First Architecture
Most edtech platforms assume a stable internet connection. Our users are electricians in basements, crane operators at remote sites, and HVAC technicians in mechanical rooms. VoltExam runs all learning algorithms, question selection, and progress tracking entirely on-device.
On-device inference — The adaptive engine runs natively on iOS (Swift/CoreML) and Android (Kotlin/ML Kit). No server round-trips for question selection, meaning zero latency even with no connectivity.
Lightweight models — Our proficiency models are designed to run within the memory and compute constraints of mobile devices. The full question bank and learning model for any trade fits within a single app download.
Sync when available — When connectivity is restored, anonymized learning data syncs to improve global difficulty calibration and question quality — creating a feedback loop that makes the system smarter for every user.
Data Advantage
With 34 live apps spanning 34+ trade categories and over 22,000 practice questions, VoltExam is building the largest proprietary dataset of trade certification learning patterns in the market. This data powers continuous improvement across three dimensions.
Question quality scoring — We measure discriminative power for every question: does getting it right actually predict exam readiness? Low-signal questions get flagged for revision or replacement.
Cross-trade transfer learning — Many trades share foundational knowledge (electrical safety, OSHA standards, blueprint reading). Our models identify these overlaps and can bootstrap proficiency estimates for new trades based on performance in related ones.
Cohort benchmarking — Anonymized aggregate data lets us tell a user not just their absolute proficiency, but how they compare to other learners who went on to pass (or fail) the same exam.
Integrated Trade Calculators
Every VoltExam app includes domain-specific calculators that are useful both during exam prep and on the job afterward. This is a deliberate product decision: when the app stays useful post-exam, it drives organic referrals from colleagues on the job site.
12 calculators are also available free on the web at voltexam.com/tools, serving as top-of-funnel discovery tools. These include Ohm's Law, conduit fill, voltage drop, refrigerant charge, crane load capacity, and more — each built to NEC/OSHA/trade-specific standards.
What We're Building Next
AI-generated explanations — Context-aware explanations for wrong answers, drawing from trade manuals and code references specific to the question domain. Not generic AI — trained on actual trade certification material.
Personalized study plans — Given a target exam date and current proficiency, the engine will generate a day-by-day study schedule that maximizes pass probability within the available time.
Employer dashboards — B2B analytics for trade schools and employers to track cohort progress, identify at-risk learners, and measure training ROI across their workforce.
Natural language search — Ask a question in plain English ("What's the minimum wire gauge for a 30-amp circuit?") and get the relevant practice questions, calculator, and code reference — all from within the app.
Technical Foundation
iOS — Swift, SwiftUI, CoreML, Core Data. Native performance, on-device ML inference.
Android — Kotlin, Jetpack Compose, ML Kit, Room. Matching feature parity with iOS.
Web — Next.js, TypeScript, Tailwind CSS. SEO-optimized for trade certification discovery.
Infrastructure — Vercel (web), Supabase (auth, sync), Stripe (payments). Designed for low operational overhead at scale.
Content pipeline — Proprietary tooling for rapid question authoring, validation, and deployment across all 34 apps simultaneously.
Interested in Our Technology?
We're actively exploring partnerships with trade schools, workforce development programs, and organizations interested in AI-powered certification training.
Reach out at arun@voltexam.com