How zaarakhan-dev/shiksha-sathi works
Shiksha Sathi is a competency-diagnosis service for statistical-officer training cohorts, not a course platform. Trainers upload rosters or enter assessment scores; the backend combines quiz and appraisal signals, diagnoses prerequisite bottlenecks, and returns an ordered remediation path. Trainers can record reassessment results and inspect cohort bottlenecks. The README also describes a prototype dashboard and human-reviewed AI question workflow. Implementation evidence is concentrated in the two excerpted core files; links for unsampled UI and question-generation behavior are marked unverified or omitted.
Main entry points: main.py Backend API · prerequisite_engine.py Bottleneck diagnosis · frontend Trainer dashboard · app AI question generation
Components 17
Cohort setup
- Roster upload
backend/app/main.py - Cohort records
backend/app/main.py - Assessment scores
backend/app/main.py - Backend API
backend/app/main.py
Scoring and diagnosis
- Signal scoring
ml/prerequisite_engine.py - Prerequisite DAG
ml/prerequisite_engine.py - Bottleneck diagnosis
ml/prerequisite_engine.py - Coverage and confidence
ml/prerequisite_engine.py
Learning and review
- Ordered remediation path
ml/prerequisite_engine.py - AI question generation
backend/app - Trainer review queue
backend/app/main.py
Progress and insight
- Trainer dashboard
frontend - Reassessment comparison
backend/app/main.py - Cohort heatmap
backend/app/main.py
Other components
- Trainer
- Learner
- Self-hosted local LLM
Connections 22
- Trainer → Trainer dashboard: uses
- Trainer → Roster upload: uploads roster
README.md - Roster upload → Cohort records: stores officials
backend/app/main.py - Trainer → Assessment scores: enters scores
README.md - Assessment scores → Cohort records: updates scores
backend/app/main.py - Trainer dashboard → Backend API: requests results
- Backend API → Cohort records: reads official
backend/app/main.py - Backend API → Signal scoring: scores signals
backend/app/main.py - Signal scoring → Bottleneck diagnosis: provides scores
ml/prerequisite_engine.py - Bottleneck diagnosis → Prerequisite DAG: traverses
ml/prerequisite_engine.py - Backend API → Bottleneck diagnosis: runs diagnosis
backend/app/main.py - Backend API → Coverage and confidence: reports coverage
backend/app/main.py - Backend API → Ordered remediation path: builds path
backend/app/main.py - Ordered remediation path → Learner: guides remediation
README.md - Self-hosted local LLM → AI question generation: drafts questions
README.md - AI question generation → Trainer review queue: submits for review
README.md - Trainer → Trainer review queue: approves questions
README.md - Learner → Reassessment comparison: takes retest
README.md - Backend API → Reassessment comparison: records results
backend/app/main.py - Reassessment comparison → Cohort records: updates scores
backend/app/main.py - Backend API → Cohort heatmap: aggregates cohort
backend/app/main.py - Cohort heatmap → Trainer dashboard: presents bottlenecks