06 — Technology Stack
Architecture Principle
Platform-first: every process runs through software. No Excel-based operations. This is how we maintain quality at scale and avoid the PMKVY trap of unmonitored, unverifiable delivery.
Student-Facing Systems
| System | Purpose | Build vs Buy | Est. Cost (Year 1) |
|---|---|---|---|
| Learning App (PWA + Android) | Course access, video lessons, assignments, progress tracking | Build (core differentiator) | Rs 40–60L |
| AI Tutor | 24/7 doubt resolution, practice generation, revision support | Build (fine-tuned LLM + RAG) | Rs 25–40L |
| Assessment Platform | Competency tests, project submissions, proctored exams, digital badges | Build (integrated with LMS) | Rs 15–25L |
| Career Portal | Job listings, profile builder, interview prep, alumni network | Build (Year 2) | Rs 10–15L |
Learning App Key Features
- Offline-first: downloads content, syncs progress when connected
- Video player with variable speed, vernacular subtitles, bookmarking
- Lab environments: cloud sandboxes (AWS/Azure), code editors, simulation tools
- Gamification: streaks, leaderboards, peer challenges
- WhatsApp integration for notifications and AI tutor access
Operations Systems
| System | Purpose | Build vs Buy | Recommendation |
|---|---|---|---|
| LMS (Learning Management) | Curriculum delivery, scheduling, trainer tools, attendance | Build (core to our model) | Custom — must integrate deeply with AI tutor and assessment |
| CRM | Student acquisition funnel, lead management, counseling workflow | Buy | Zoho CRM or LeadSquared (Rs 3–5L/year) |
| ERP | Finance, HR, procurement, compliance | Buy | Zoho One or ERPNext (Rs 2–4L/year) |
| Placement System | Employer portal, matching algorithm, interview scheduling, tracking | Build | Custom — critical differentiator |
| Analytics Dashboard | Real-time center KPIs, student progress, trainer performance, financial metrics | Build | Custom — built on open-source BI (Metabase/Superset) |
| Attendance & Biometric | Aadhaar-linked attendance (mandatory for govt funding) | Buy + Integrate | Mantra/Startek devices + custom API (Rs 1–2L) |
Content Engine
| Component | Tool/Approach | Cost |
|---|---|---|
| Video Production | In-house studio (1 room in Hub): 4K camera, teleprompter, lighting, green screen | Rs 8–12L setup |
| Screen Recording | OBS Studio (free) + professional mic setup | Rs 50K |
| Editing | DaVinci Resolve (free) + 1 full-time editor | Rs 30K/month salary |
| AI Dubbing | ElevenLabs / Murf.ai / open-source TTS for 8–10 Indian languages | Rs 3–5L/year |
| Subtitling | AI-generated (Whisper) + human review | Rs 1–2L/year |
| Localization Workflow | English master → AI translation → human review → AI dubbing → QA | 2–3 weeks per language per course |
| Content Versioning | Git-based curriculum management (Markdown + media assets) | Free (internal tooling) |
Content production target: 1 full course (40–60 hrs video) per quarter. 8–10 language versions within 1 month of English release.
Data Layer
| Capability | Method | Impact |
|---|---|---|
| Skill Profiling | Pre-assessment + learning behavior + project performance → dynamic skill graph per student | Personalized learning paths, better employer matching |
| Demand Forecasting | Job portal scraping + employer surveys + govt data → role-level demand by geography | Curriculum prioritization, center location decisions |
| Dropout Prediction | Attendance + engagement + assessment patterns → risk score (updated daily) | Early intervention: counselor call when risk > 60% |
| Quality Scoring | Trainer observation data + student outcomes + audit results → center quality index | Resource allocation, trainer coaching, public transparency |
| Placement Matching | Student skill profile + employer requirements → ranked match score | Faster, better-fit placements |
Build vs Buy Decision Framework
| Criteria | Build | Buy |
|---|---|---|
| Core differentiator? | Yes → Build | No → Buy |
| Does it need deep integration with our data? | Yes → Build | No → Buy |
| Does an affordable, adequate solution exist? | No → Build | Yes → Buy |
| Will we need to iterate rapidly? | Yes → Build | No → Buy |
| Regulatory/compliance-specific? | Maybe → Evaluate | Often → Buy certified |
Tech Team (Year 1)
| Role | Count | Monthly Cost |
|---|---|---|
| VP Engineering | 1 | Rs 2.5–4L |
| Full-stack Engineers | 3 | Rs 1–1.5L each |
| AI/ML Engineer | 1 | Rs 1.5–2.5L |
| Mobile Developer | 1 | Rs 1–1.5L |
| DevOps / Cloud | 1 | Rs 80K–1.2L |
| UI/UX Designer | 1 | Rs 60K–1L |
| QA Engineer | 1 | Rs 50K–80K |
| Total | 9 | Rs 10–15L/month |
Estimated Tech Investment
| Category | Year 1 | Year 2 |
|---|---|---|
| Product development (team) | Rs 1.2–1.8Cr | Rs 1.5–2.5Cr |
| Cloud infrastructure (AWS/Azure) | Rs 15–25L | Rs 30–50L |
| SaaS subscriptions (CRM, ERP, tools) | Rs 8–12L | Rs 12–18L |
| Content production (studio + team) | Rs 20–30L | Rs 25–35L |
| AI/LLM costs (API + fine-tuning) | Rs 10–20L | Rs 20–40L |
| Hardware (biometric, lab equipment) | Rs 8–12L | Rs 15–25L |
| Total | Rs 1.8–2.8Cr | Rs 2.5–4.2Cr |