How QikSkill works
A detailed walkthrough of how candidates, Training Partners, and administrators use QikSkill — and how the platform fits inside the NSDC / NSQF ecosystem.
Who uses QikSkill
Three distinct roles, each with its own portal, permissions, and data view.
Candidate
A PMKVY, NAPS, or self-enrolled learner. Studies NOS-mapped modules, chats with the AI tutor in their preferred language, attempts mock assessments, and tracks readiness for the formal NCVET-approved assessment.
Training Partner
An NSDC-empanelled organisation accredited via SMART. Creates batches against specific QPs, enrols candidates by email, monitors real-time analytics, and identifies at-risk learners before assessment day.
Platform admin
QikSkill's internal operations team. Approves new TP accounts, uploads Qualification Packs and NOS content, triggers embedding and chunking for the RAG pipeline, and monitors platform-wide metrics.
The candidate journey
From batch enrolment to Rozgar Mela — six stages over roughly 300 to 600 training hours, depending on the QP's NSQF level.
1 · Enrolled by TP
Your Training Partner adds your Google email, phone, name, and preferred language (EN / HI / TA) to their batch in SDMS and in QikSkill. You do not set a password.
2 · First sign-in
You go to the QikSkill login, click Continue with Google, and the platform automatically links your Google identity to the candidate record your TP already created. Your preferred language loads.
3 · NOS-mapped study
Your dashboard shows every NOS in your QP with a progress bar. Each module opens a reader view with content chunks in your language, a mic button to "Listen" via Sarvam TTS, and a one-click jump to the AI tutor.
4 · AI tutor (text + voice)
Ask anything about the current NOS. The tutor retrieves the top 4 relevant content chunks from the vector database, replies in 2–4 sentences, and attaches a curated YouTube video when your question implies a demonstration. Responses are spoken back via Bulbul v3 TTS.
5 · Mock assessment
A timed MCQ run — one minute per question, no back-navigation. Your score updates `candidate_readiness` via a Postgres RPC, weak and strong NOS areas are flagged, and every wrong answer gets a 2-sentence revision note generated by sarvam-m.
6 · Placement preparation
A voice-enabled mock interview for your specific job role. The AI asks six progressive questions, rates each answer 1–5, and upserts a placement readiness score. When you clear the formal NCVET assessment, your TP connects you to an employer at the next Rozgar Mela.
The Training Partner workflow
1 · Sign up and onboard
Continue with Google and complete a 30-second onboarding: organisation name, NSDC TP code, optional phone. The account is created with status = pending until the platform admin approves it.
2 · Create a batch
From the QP library, pick the Qualification Pack your TP is empanelled for. Set start and assessment dates aligned to your PMKVY batch calendar, target enrolment count, and the batch is live.
3 · Enrol candidates
Add candidates one-by-one with name, Gmail, phone, and preferred language. No passwords to manage. On the candidate's first Google sign-in, QikSkill matches by email and links the auth identity automatically.
4 · Monitor analytics
The Analytics page aggregates real data — no self-reports. Four live metrics (active learners, average readiness, pass rate, voice-session %), a 30-day trend chart, a per-NOS pass-rate bar chart, and an at-risk list of candidates inactive for 7+ days.
5 · Generate reports
Every learning session, chat message, and assessment attempt is INSERT-only and timestamped. At scheme audit time (SSC, NCVET, CAG) you export raw logs with cryptographic integrity guarantees — no manual reconstruction required.
The platform admin workflow
1 · Approve TPs
New training partners arrive with status = pending. The admin verifies the NSDC TP code against SMART, then flips status to active or suspended via the admin TPs table.
2 · Upload QP content
Add a new Qualification Pack from the admin QPs page, then add each NOS unit under it. Paste model-curriculum text in all three languages; QikSkill chunks on blank lines and calls HuggingFace multilingual embeddings inline so the RAG retrieval works the moment a candidate opens the tutor.
3 · Monitor platform health
The admin dashboard aggregates everything — total TPs, QPs, candidates, learning hours, voice percentage, language distribution, top QPs by enrolment, and a 30-day session-count trend across all tenants.
Where QikSkill fits in the NSDC ecosystem
QikSkill is a learning-delivery layer — it sits between the SSC-authored curriculum and the candidate. It does not replace any official portal; it augments them.
QikSkill augments
- Candidate learning quality (adds AI tutor, voice, and practice)
- TP visibility (adds real-time analytics on actual engagement)
- Audit reliability (adds tamper-proof evidence for every learning minute)
QikSkill does NOT replace
- SDMS · official batch registration and certification records
- SMART · TP accreditation and training-centre grading
- Takshashila · trainer and assessor registration
- NCVET-approved Assessment Agency · final third-party assessment
- Skill India Digital Hub · official course directory and LMS
What data lives where
Total transparency about every piece of information QikSkill stores and who can see it.
| Data | Stored | Visible to |
|---|---|---|
| Learning sessions (start / end / duration) | Supabase Postgres · INSERT-only | The candidate + their TP + admin |
| Chat messages with the AI tutor | Supabase Postgres · INSERT-only | The candidate only (private) |
| Assessment attempts (score, language used) | Supabase Postgres · INSERT-only | The candidate + their TP + admin |
| Readiness scores (aggregated) | Supabase Postgres · recalculated via RPC | The candidate + their TP |
| Audio clips (speech-to-text input) | Not persisted · sent once to Sarvam, then discarded | Never stored |
| YouTube resources | Cached 7 days in nos_youtube_resources | All candidates studying that NOS |
Assessment and certification
QikSkill's mock assessments are formative — they prepare the candidate for the real thing. They are not certifications and they do not replace NCVET-approved Assessment Agencies.
The formal PMKVY / NSQF certification is issued by an independent third-party Assessment Agency empanelled with NCVET, acting on behalf of the relevant Sector Skill Council. A Training Partner cannot self-assess. This separation is deliberate — the post-2025 CAG audit reforms specifically reinforced it.
QikSkill's role: make sure the candidate walks into the formal assessment already ready. Readiness score ≥ 70 correlates strongly with passing the real exam.
Placement and the Rozgar Mela link
Once a candidate is certified, their TP has a statutory obligation to provide placement assistance. QikSkill's Placement Prep module is an AI-powered mock-interview trainer for the candidate's specific job role.
Candidates who complete mock interviews arrive at the Rozgar Mela (organised every six months under PMKVY) with an objective placement score derived from AI feedback on clarity, relevance, and improvement — data their TP can share with prospective employers.
For sector-specific roles, certified candidates can also plug into NSDC JobX / KaushalSangam — the digital placement platform that matches PMKVY-certified talent to vacancies nationally.
Frequently asked questions
Launch checklist for a new Training Partner
Five things to have in place before you run your first QikSkill-backed batch.
- 1Active NSDC TP code and SMART accreditation for the sectors you intend to train in
- 2A nominated super-admin email (we recommend a distribution list, not a personal address)
- 3A list of candidate Google emails ready to bulk enrol (any Gmail works; corporate Google Workspace too)
- 4An assessment date aligned with NCVET-approved Assessment Agency availability
- 5A clear communication to candidates about language preference (EN / HI / TA) at the time of enrolment
Questions we didn't answer here?
Reach the team through the contact page below. We reply within two business days on standard inquiries.
Contact us →