eLearning and education software development

Education software development for schools, universities and edtech

Education software has more users with different rights than most products: students, teachers, parents, school admins and sometimes a district or a university office above them. Getting roles and data boundaries right is most of the work. For Basira we built four education products over a multi-year engagement, all four now live: GyaanSchool for K-12, Qoollege for higher education, PiSolved for AI-assisted problem solving, and Sequent, an LLM RAG platform for one of Basira's enterprise clients. Each had a different stack, from .NET and Angular to Next.js and Python.

eLearning software development, from MVP to multi-product platform

Most eLearning platforms start as one course flow and grow into many services. We plan for that growth without building it on day one. A first version covers sign-in, roles, content and one learning loop. Later, the pieces that grow fastest, such as questions, video, search and notes, become their own services. GyaanSchool runs that way: a .NET GraphQL core, Node.js services for authentication, questions, search, videos, notes and background work, and Angular front ends, shipped through Azure DevOps pipelines. Qoollege runs on Next.js on its own Kubernetes cluster, with a pipeline that gathers university data into the platform.

Custom LMS development: when to build and when to extend

A learning management system (LMS) handles courses, enrolments, assignments, grades and reports. For standard courses, an existing LMS such as Moodle or Canvas is often enough, and we will say so. Custom LMS development makes sense when your teaching model does not fit: speaking practice, adaptive question banks, a marketplace of tutors, or reporting your schools actually ask for. A middle path is common too: keep the LMS for enrolment and grades, and build the custom learning experience beside it, connected through its API or the LTI standard.

AI tutors and speaking practice

AI tutoring works when the tutor responds fast enough to feel like a conversation. For Lang AI Tutor, we built AI-avatar language tutors on a realtime pipeline: speech to text, a language model, then text to speech, at conversational latency, with a video avatar so learners talk to a consistent persona instead of a chat box. The tutor keeps context across turns, so it can refer back to what the learner said and adjust the practice.

We are also building our own AI speaking tutor for schools, at proof-of-concept stage: a teacher assigns a timed speaking session, the student practises with an avatar, and both see a formative score for fluency, accuracy and vocabulary. It stores the assessment only, not raw audio or video.

Assessment and screening tools

Assessment software must support professional judgement, not replace it. For Noetic we built a screening platform covering ADHD, autism, dyslexia, dyspraxia and dyscalculia, with a structured intake and reports that help clinicians and learning-support staff make the final call. When we add AI to assessment, it produces formative feedback with the reasoning shown, and a teacher or specialist owns the grade.

Built for FERPA-regulated data, COPPA and GDPR

Student data carries legal duties that differ by country and by age. We build to the controls your school, district or university sets:

• Access limited by role, school and class, so a teacher sees only their own students
• Data minimisation: collect what the learning flow needs and nothing more
• Parental consent flows and restricted features for users under 13, where COPPA applies
• EU or UK hosting for GDPR, and deletion on request
• Logs of who viewed or changed student records

Sign-off stays with your institution. We document how each control is met so your data protection lead can review it.

EdTech software development for startups

EdTech founders need to prove that schools will use and pay for the product before they build everything. We build an MVP from $5,000: one learning flow, accounts and roles, and a deployment you own. When the product works, we stay on as a monthly team and set up the infrastructure to scale it, from CI/CD pipelines to Kubernetes (see DevOps and cloud).

What education software costs

An eLearning MVP starts from $5,000 for one core learning flow with accounts, roles and a deployment you own. AI tutoring, speech, LMS integrations and multi-school reporting add to that, and AI features start from $8,000. Model and cloud costs are billed to your own accounts. You get a fixed quote in writing after a short discovery. See pricing.