AI-Driven voice Assistant for livelihood Mapping and NSQF-Aligned Skilling Recommendations for SC Communities under GIA component of PM-AJAY
Ministry of Social Justice and Empowerment (MoSJE) · Smart Education · Software
The published qualification framework gives you the real taxonomy that every other skill-matching statement here is missing, and a live phone call in a regional language is a demo nobody else will have — build the IVR path first, because that is the accessibility claim and the differentiator at once.
What it actually is
People are enrolled in skill training programmes that do not match what they can do, want to do, or what there is work for locally, so they drop out or finish and find nothing. The forms that would capture that mismatch are text-heavy and in the wrong language. The ask is an assistant that has a spoken conversation instead, in the person's own dialect, and recommends training that actually fits.
What to build
A voice-first profiling and recommendation assistant that conducts a conversational interview rather than a form, collecting the seven areas the statement enumerates — education, traditional family occupation, current livelihood, skills and interests, mobility constraints, self-employment versus wage preference, and local conditions — in regional languages and dialects, then maps that profile against the published national qualification framework so recommendations resolve to real, named, certifiable job roles rather than vague trade suggestions, identifies the specific competency gaps between where the person is and the entry requirements of each recommended qualification, and reaches users where they are through an IVR path that works on a feature phone with no app, no data connection and no literacy requirement.
Smallest thing that wins the room
Call a phone number from an ordinary handset, hold a two-minute conversation in a regional language, and receive back three named qualification-framework job roles with the specific gaps between the caller's stated background and each one's entry requirements.
How crowded this one gets
A guess, projected from the 2025 statements — the last year where both the submission counts and the winners were published.
Quieter than 29% of the 226 · #160 of 226 by expected field
A normal-sized field. Your idea has to be good, not miraculous.
Why: central ministry statements sat below the average.
This is a guess, not a fact
Nobody has published 2026’s numbers yet. This is an analysed estimate from last year’s pattern, so please do not take it as the truth — check the live counter on the SIH portal before you decide anything. The range covers the middle half of likely outcomes, so one statement in two lands outside it. Entry closes at 500 ideas per statement, so no range goes past that — a statement that reaches the cap fills and shuts rather than drawing an unlimited crowd. The model reads only three things a team can see before choosing — software or hardware, the theme, and what kind of body posted it — and those explain about a quarter of the variation in last year’s field sizes (R² 0.25 on held-out statements). Trust the band more than the number, and the ordering more than either. It cannot see how good your idea is, which is the part that actually decides it.
The scores
The number is the shorthand. The line under it is the reason.
Acceptance potential
4/5The national qualification framework gives you a real published competency taxonomy to map against, which is exactly what the skill-matching statements elsewhere on this portal lack, the feature-phone channel is a genuine accessibility decision almost nobody will build, and a live phone call is a demo no competing team will have.
Feasibility
4/5The pieces are all available and free — national language infrastructure provides speech recognition and synthesis across Indian languages, the qualification framework publishes its job roles and competency requirements as a real taxonomy you can map to, and telephony providers make IVR straightforward — with only local labour demand at block level being genuinely unavailable.
Innovation scope
3/5The conversational approach, the information to collect and the deployment channels are specified, but how you conduct an adaptive interview that feels empathetic rather than administrative, and how you map informal spoken experience onto formal qualification descriptors, are open and are the real problems.
Clarity
4/5The information to collect, the outputs, the accessibility constraints and the deployment channels including feature-phone IVR are all specified clearly, though the section listing basic issues under the programme is a set of administrative grievances that has nothing to do with the solution being asked for.
Effort
HeavyA dialogue manager for an adaptive spoken interview, multilingual speech in and out, profile extraction from unstructured speech, qualification mapping and gap analysis, and IVR plus messaging deployment is five workstreams, with the dialect coverage being the most open-ended.
Demo-ability
EasyA judge picks up a phone, dials a number and has a conversation in a regional language that produces a real recommendation — a live phone call is an unusual and memorable demo format, and it needs no screen at all.
In its favour
- Green flag: The national qualification framework is a real, published, structured taxonomy of job roles with defined competencies, so your recommendations resolve to certifiable qualifications rather than to categories you invented — this is precisely what the other skill-matching statements on this portal cannot do
- Green flag: The feature-phone IVR path is the single most important design decision here and almost no team will build it: the people this serves may not have a smartphone, and a team that demos on a real phone call has answered the accessibility requirement rather than described it
- Green flag: Speech across Indian languages is free through national language infrastructure, so the hardest-sounding requirement has a supported answer
- Green flag: Asking about traditional family occupation and mobility constraints is unusually thoughtful profiling, and building the interview to surface those naturally rather than as form fields is a genuine design contribution
Against it
- Red flag: Local labour market demand at block level does not exist as data, so the region-specific opportunity half of the recommendation has to come from qualification-framework sector mapping and stated local context rather than from real vacancy data — be clear which
- Red flag: Dialects are where speech recognition fails, and the users least served by existing systems are exactly the ones whose speech is furthest from the training data, so test on genuine dialect speech rather than on standard-register recordings
- Red flag: An empathetic conversational interview that quietly becomes an interrogation is a real risk — the person is disclosing their circumstances to a government system, so keep the interview short and let them decline questions
- Red flag: Recommending a training programme that does not lead to work reproduces the exact failure the statement describes, so present the gap analysis honestly rather than optimistically matching everyone to something
What you will be writing
- Bhashini ASR and TTS across regional dialects
- NSQF and NCO job role taxonomy mapping
- IVR telephony for feature phone access
- adaptive spoken interview dialogue manager
- competency gap analysis against entry requirements
- WhatsApp voice note interface
- Skilling and livelihood promotion
- Voice interfaces for low-literacy users
- Social welfare programme delivery
Prior art to read before you start
conversational profiling instead of form filling · NSQF-aligned training recommendation · IVR delivery for low-connectivity users
Analysed by Claude Opus. Every score above is a judgment call with its reasoning attached — kindly cross-check this against the official statement on the SIH portal before your team commits to it.