How adithyap-72/career_guidance_ml_ops works
This is an Indian career-guidance application backed by an MLOps workflow. A job seeker supplies skills and experience (or uploads a resume); the serving API returns ranked roles, readiness and skill gaps, and course suggestions. The repository also covers preparing labeled job-posting data, training and evaluating role classifiers, and packaging a model for inference. README documents the product outcomes; source excerpts and the SOURCE INDEX support only some internal calls, so uncertain links are omitted.
Main entry points: app.py Inference API · recommender.py Role recommender · train.py Model training · bundle.py Model bundle
Components 25
Career guidance
- Web dashboard
frontend/app.js - Course matching
src/models/courses.py - Role skill profiles
src/models/profiles.py
Data preparation
- Source snapshots
data/raw - Posting ingestion
src/data/ingest.py - Role labeling
src/data/label.py - Role taxonomy
src/data/taxonomy.py - Data profiling
src/data/profile.py - Dataset splitting
src/data/split.py
Model lifecycle
- Feature pipeline
src/features/build.py - Role estimators
src/models/estimators.py - Model training
src/models/train.py - Model evaluation
src/models/evaluate.py - Experiment tracking
src/models/tracking.py - Champion selection
src/models/select.py - Quality gate
src/models/validate.py - Model registry
src/models/register.py
Inference
- Inference API
src/serving/app.py - Resume extraction
src/serving/resume.py - Role recommender
src/serving/recommender.py - API schemas
src/serving/schemas.py - Request limits
src/serving/limits.py - Skill keyness
src/serving/keyness.py - Model bundle
src/serving/bundle.py
Other components
- Job seeker
Connections 28
- Job seeker → Inference API: submits profile
README.md - Inference API → Resume extraction: extracts resume
src/serving/app.py - Inference API → Role recommender: requests recommendations
src/serving/app.py - Inference API → Request limits: rate limits
src/serving/app.py - Inference API → API schemas: uses schemas
src/serving/app.py - Model bundle → Role recommender: provides model files
src/serving/recommender.py - Role recommender → Course matching: matches courses
src/serving/recommender.py - Role recommender → Role skill profiles: reads role skills
src/serving/recommender.py - Job seeker → Web dashboard: uses interface
- Web dashboard → Inference API: calls API
- Source snapshots → Posting ingestion: loads snapshots
src/data/ingest.py - Posting ingestion → Role labeling: prepares postings
- Role labeling → Role taxonomy: assigns roles
- Role taxonomy → Dataset splitting: supports role data
- Data profiling → Role skill profiles: builds profiles
- Dataset splitting → Feature pipeline: provides training data
src/data/split.py - Model training → Feature pipeline: builds inputs
src/models/train.py - Model training → Role estimators: builds classifier
src/models/train.py - Model training → Model evaluation: scores model
src/models/train.py - Model training → Experiment tracking: logs run
src/models/train.py - Champion selection → Model evaluation: compares metrics
src/models/select.py - Champion selection → Experiment tracking: uses run records
src/models/select.py - Quality gate → Model evaluation: checks performance
src/models/validate.py - Model registry → Experiment tracking: uses run metadata
src/models/register.py - Model registry → Model bundle: supplies registered model
src/models/register.py - Model bundle → Model training: checks trusted types
src/serving/bundle.py - Role recommender → Feature pipeline: uses model preprocessing
src/serving/recommender.py - Role recommender → Skill keyness: uses skill lookup