Finalist defence · Evidence-bounded
Seven questions judges should ask us
Short answers backed by behavior in the working synthetic prototype—not promises, invented impact, or implied government approval.
Bharat Career ID is a source-aware compatibility layer between verified credentials and different recruitment forms.
The receiving question determines the authority. The citizen sees the reason, approves purpose-specific sharing, and keeps a record of exactly what crossed the simulated boundary.
1Is this just another OTR?
No. An agency OTR reuses data inside one system. Bharat Career ID demonstrates a compatibility layer across different receiving schemas, where the same citizen can need different sourced values for differently worded fields.
Working proof: Open the two-form path: Matriculation name and current legal name resolve differently.2Why not DigiLocker plus autofill?
A document wallet supplies records, while this prototype tests a narrower decision: which sourced value a particular receiving field supports. It makes that policy visible, stops on missing or unsuitable evidence, and asks the citizen before reuse.
Working proof: The resolver shows field → authority rule → credential/value → citizen decision.3How do you protect privacy?
The prototype keeps progress encrypted in this browser, records purpose-specific consent for exact values and versions, supports decline and revocation, and creates no real government or credential connection.
Working proof: Inspect the consent receipt, local event ledger, and immutable Application Passport.4Can AI make a wrong eligibility decision?
AI may propose cited rule structures, but it cannot publish them or evaluate a citizen. A reviewer approves validated rules; deterministic code evaluates them; unclear evidence fails closed for review.
Working proof: The reviewer mutation replay changes a 60% rule to 90% and the same typed rule changes the citizen result.5How would government adoption start?
With one recruitment body, one approved credential source, and two high-risk fields. The pilot would agree a compatibility contract for field meaning, accepted authority, consent, acknowledgement, security, and status—not replace the body's portal.
Working proof: The portability audit separates mapped terminology from unsupported OTP, payment, biometrics, and account behavior.6Where is the impact evidence without real participants?
There is no claim about real-world error reduction or comprehension. The evidence is software behavior: labelled synthetic cases, adversarial failures, desktop/mobile journeys, and one captured payload where the resolver prevents a demonstrable source mismatch.
Working proof: Human participants remain 0. Every citizen and outcome is labelled synthetic.7Why should judges select Bharat Career ID for the top 10?
It turns a subtle source-authority failure into a working, inspectable citizen journey. Two receiving schemas resolve the same citizen's name differently for a legitimate reason; missing evidence stops one field without breaking the other; purpose-limited consent and a preserved local handoff record close the loop. AI proposes rules, a reviewer approves them, and deterministic code evaluates them.
Working proof: Run the authority replay, complete the citizen path, then inspect the labelled regression evidence and explicit limits.