Unified Career Infrastructure

NEP 2020, the mandate we build against

The National Education Policy 2020 sets the direction; fourteen of its commitments define the infrastructure this platform is designed to serve. Enrolment and platform figures per Ministry of Education, 2023-24.

• Academic Bank of Credits Every learner's credits are to be digitally stored, transferred and redeemed across institutions, making the academic record portable for the first time.
• Multiple entry and exit Undergraduate education is to allow exit after one, two, three or four years with certificate, diploma or degree, requiring records that persist across interruptions.
• One curricular continuum The 10+2 structure is replaced by 5+3+3+4, covering ages 3 to 18, so the learning record begins earlier and runs longer.
• Universal foundational literacy and numeracy Every child is to attain foundational literacy and numeracy, tracked from the earliest grades.
• 100% school enrolment A Gross Enrolment Ratio of 100% in school education is targeted by 2030, which demands finding and tracking every out-of-school child.
• 50% higher-education enrolment GER in higher education is to reach 50% by 2035; it has already moved from 23.7% to 30% in a decade, with total enrolment up from 3.42 to 4.50 crore.
• Skill integration with NCrF Vocational and academic learning are unified under the National Higher Education Qualifications Framework aligned with the National Credit Framework, so skill credits and degree credits live in one system.
• Apprenticeship-embedded degrees Degree programmes are to embed internships and apprenticeships (IEDP, NATS), putting verified work experience inside the academic record itself.
• PARAKH A national assessment centre is to set norms and standards for student assessment across boards, implying comparable records nationwide.
• Digital learning at national scale Platforms like SWAYAM Plus carry 500+ courses with 89 industry partners across 17 sectors, with six lakh learners already registered, each generating credentials to be recognised.
• National Educational Technology Forum A dedicated platform is to guide how technology is deployed across education, from learning to governance.
• Continuous teacher development Every teacher is expected to complete at least 50 hours of professional development annually, itself a record to be maintained.
• Equity for disadvantaged groups Female GER has risen from 22.9% to 31.2%, SC from 18.9% to 27.8%, ST from 13.5% to 22.8% over the decade, gains that depend on identifying and following each learner.
• Lifelong learning Education is framed as continuing through life, beyond formal degrees, so the record can never be considered closed.

One architecture, one citizen, every system

A proprietary modelling stack is envisioned to be operated on sovereign, in-country data centres, so that the longitudinal citizen record could be turned into predictive national infrastructure.

• Longitudinal modelling core Temporal graph neural networks and survival models are envisioned to be trained over each citizen's education-to-employment sequence, so that trajectories could be forecast through sequence transformers and Bayesian state-space models rather than described after the fact.
• Policy-grounded LLM layer A domain-tuned language model, aligned through retrieval-augmented generation over NEP 2020 and the DPDP Act and constrained by constitutional-AI guardrails, is envisioned to be deployed, so that inference, retention and consent could be bounded by policy encoded as the model's operating logic.
• Agentic orchestration A mesh of autonomous AI agents, for eligibility, credential verification, skill-gap detection and placement matching, is envisioned to be coordinated through a multi-agent planner over a shared feature store, so that cross-departmental workflows could be executed with a human-in-the-loop.
• Demand-supply intelligence Real-time labour-market signals are envisioned to be fused with the skills graph through recommender systems and causal-inference models, so that structural mismatches could be predicted by digital-twin simulation of the workforce before they surface.
• Sovereign MLOps and benchmarking A federated, privacy-preserving training pipeline with differential privacy and continuous evaluation is envisioned to be run on proprietary GPU infrastructure, so that workforce readiness could be scored and benchmarked against forward demand rather than held credentials.

Together, this stack is envisioned to converge on a Personalized Career Prescription, a per-citizen, continuously updated recommendation of the next qualification, skill and role, computed against where the economy is heading.