// Curriculum Highlights
What You'll Learn
Docker & ContainerisationPackage ML models into reproducible containers — runs anywhere, same result every time
FastAPI Model ServingBuild production REST APIs for ML models with validation, error handling, async endpoints
Experiment Tracking with MLflowLog parameters, metrics, artifacts — never lose a run, always reproduce your best model
CI/CD for ML PipelinesGitHub Actions workflows: test → train → evaluate → containerise → deploy automatically
Kubernetes Basics for MLPods, deployments, services, HPA — run and scale ML services on K8s clusters
Model MonitoringEvidently AI for data drift, concept drift, and performance degradation detection
Automated RetrainingTrigger retraining on drift detection, schedule pipelines with Airflow or Prefect
Cloud Deployment (AWS/GCP)EC2, S3, ECR, SageMaker basics — deploy models to the cloud and serve at scale
Feature StoresFeast for managing features consistently between training and serving — eliminate training-serving skew
ML System DesignDesign production ML systems: data pipelines, model registry, A/B testing, rollback strategies
// The Full Picture
ML in Production — The Complete Lifecycle
Most ML courses end at model training. This course starts there. You'll own the entire lifecycle — every phase covered with real tools used by production ML teams.
Phase 1
Data
- Pipelines
- Versioning
- Feature stores
- DVC
›
Phase 2
Experiment
- MLflow tracking
- Model registry
- Hyperparameter optimisation
›
Phase 3
Package
- Docker containers
- FastAPI serving
- ONNX export
›
Phase 4
Deploy
- CI/CD pipelines
- Kubernetes
- Cloud (AWS/GCP)
›
Phase 5
Monitor
- Drift detection
- Grafana dashboards
- Auto-retraining
// Note: This lifecycle is circular — Phase 5 feeds back into Phase 1 via automated retraining triggers. That loop is what makes ML systems robust in production.
// Production Toolstack
20+ Tools You'll Actually Use
No toy examples. Every tool below is used on a real project in the course. These are the exact tools on job descriptions at ML engineering companies in India and globally.
Serving & APIs
FastAPI
Uvicorn
BentoML
Triton Inference
Tracking & Registry
MLflow
Weights & Biases
DVC
HuggingFace Hub
Infra & CI/CD
Docker
Kubernetes
GitHub Actions
Terraform
Monitoring
Evidently AI
Prometheus
Grafana
Great Expectations
// Automation
Automated ML Pipeline — Built in Week 3
By week 3 you'll have a fully automated CI/CD pipeline that handles the entire model lifecycle — from a git push to a live, monitored endpoint.
⚙️
fraud-detection-pipeline · main branch
All checks passing
Code
Tests
Tests
Data
Validation
Validation
Model
Training
Training
Evaluate
& Gate
& Gate
Build
Container
Container
Deploy
Staging
Staging
Smoke
Test
Test
Deploy
Prod
Prod
✓ Evaluation gate: if new model F1 < current production F1 − 0.01, deployment is blocked. Zero regressions reach production.
// After This Course
Career Outcomes
MLOps Engineer
Dedicated MLOps roles at companies with mature ML infrastructure — handling pipelines, monitoring, and reliability.
₹15–30 LPA rangeAI Platform Engineer
Build the internal tooling and infrastructure that other ML engineers depend on — high leverage, high impact roles.
₹18–35 LPA rangeFull-Stack ML Engineer
The rarest and most valuable profile: an engineer who can build, train, deploy, monitor, and scale ML systems end-to-end.
₹20–40 LPA rangeCloud ML Architect
AWS/GCP ML specialists designing the cloud infrastructure for ML workloads — SageMaker, Vertex AI, GKE expertise.
₹22–45 LPA range// This course is for
🤖 ML engineers who can build models but want to learn to ship them properly
💻 Software engineers who want to bridge into ML infrastructure and DevOps
🎓 Those who completed ML Mastery and want the production-readiness layer
❌ Not for: absolute beginners — assumes Python and basic ML model knowledge
// Week by Week
Full Curriculum — 5 Weeks
- Why containers: reproducibility, environment isolation, portability
- Docker fundamentals: images, containers, Dockerfile, layers
- Multi-stage builds for slim production images
- FastAPI: routing, Pydantic models, async endpoints, middleware
- Serve a Scikit-learn model via REST API with input validation
- Docker Compose: multi-container setups (API + database + monitoring)
- Health checks, readiness probes, graceful shutdown
- Load testing with Locust: latency, throughput, error rate
- MLflow: tracking runs, logging metrics/parameters/artifacts
- MLflow Model Registry: staging → production promotion workflow
- Weights & Biases: experiment comparison, sweep visualisation
- DVC: data version control, remote storage (S3/GCS), pipeline DAGs
- Feature stores with Feast: offline/online feature retrieval
- Training-serving skew: causes, detection, prevention
- Model cards: documenting model behaviour, limitations, and bias
- ML-specific testing: unit tests for transforms, integration tests for the API
- Great Expectations: data quality validation before training
- GitHub Actions: workflow YAML, triggers, matrix builds, secrets
- Full ML CI/CD pipeline: push → test → train → evaluate → gate → build → deploy
- Evaluation gate: block deployment if metric regression detected
- Container registry: push to AWS ECR or GitHub Container Registry
- Canary deployments and blue-green strategy for ML services
- Kubernetes core: pods, deployments, services, namespaces, configmaps
- Deploy ML API to local K8s cluster (minikube)
- Horizontal Pod Autoscaler (HPA): scale on CPU/RPS
- AWS basics for ML: EC2, S3, ECR, IAM roles
- Deploy to EKS (Elastic Kubernetes Service) or GKE
- SageMaker introduction: managed training and endpoints
- Infrastructure as Code with Terraform: provision reproducible ML infra
- Model monitoring fundamentals: data drift, concept drift, performance decay
- Evidently AI: generate drift reports, data quality reports
- Prometheus: instrument FastAPI with custom ML metrics
- Grafana: dashboards for model performance, throughput, latency
- Alerting: PagerDuty/Slack integration for model health alerts
- Automated retraining: trigger on drift threshold with Airflow/Prefect
- A/B testing ML models in production: shadow mode, split traffic
- Capstone: end-to-end MLOps system for a real ML use case
// Hands-on Work
4 End-to-End Deployment Projects
PROJECT 01
Containerised Fraud Detection API
Package the fraud detection model from ML Mastery into a FastAPI service, containerise it with Docker, document the API with OpenAPI, and load-test it with Locust.
PROJECT 02
Experiment Tracking Dashboard
Set up MLflow tracking for 5 model variants on the same dataset. Build a model registry with staging/production lifecycle and demonstrate reproducible runs from DVC-versioned data.
PROJECT 03
Full CI/CD ML Pipeline
Build a GitHub Actions pipeline that automatically trains, evaluates, gates, containers, and deploys an updated model on every push to main — with Slack notifications at each stage.
PROJECT 04 — CAPSTONE
Production ML System with Monitoring
Full system: model served via FastAPI on Kubernetes, CI/CD pipeline via GitHub Actions, Evidently drift reports, Prometheus + Grafana dashboard, automated retraining trigger on drift detection.
🏗️
Architecture Templates Included
3 reusable MLOps architecture templates: Batch Prediction Service, Real-Time Inference API, and Continuous Training Pipeline. Production-ready Dockerfile, docker-compose, K8s manifests, and GitHub Actions YAML — copy, customise, deploy.
// Your Credential
Platinum Certificate Awarded
💎
Newton JEE Platinum Badge
AI Engineer — MLOps & AI Engineering
Appears on your LinkedIn profile
The Platinum Badge — Full-Stack AI Engineer
The Platinum badge is the capstone of the Newton JEE certification ladder. It signals to recruiters that you can take a model all the way from research to production — the complete engineering skill set. Combined with Silver or Gold badges, it creates one of the most comprehensive AI credential clusters visible on LinkedIn.
1
Complete all 5 weeks and 4 projects2
Deploy capstone system with live Grafana dashboard3
Present architecture walkthrough to mentor4
Platinum credential link issued within 48hrs5
One-click publish to LinkedIn Certifications// Your Mentor
Meet Your Instructor
VS
Vikram Subramaniam
Principal MLOps Engineer · Ex-Swiggy & Razorpay
10 years in software infrastructure, the last 6 focused entirely on ML systems at scale. Built Swiggy's demand forecasting pipeline (serving 500k+ predictions/hour) and Razorpay's fraud model CI/CD infrastructure from scratch. Vikram is blunt about what actually matters in MLOps: most problems aren't algorithm problems — they're infrastructure and reliability problems. This course teaches you to solve the ones that actually get engineers woken up at 2am.
// Upcoming Batches
Pick Your Batch
Batch #07
May 17, 2025
Sat & Sun · 2:00–4:00 PM IST
6 seats left
Batch #08
Jun 7, 2025
Sat & Sun · 10:00 AM–12:00 PM IST
12 seats open
Batch #09
Jun 28, 2025
Sat & Sun · 2:00–4:00 PM IST
15 seats open
// Ready to Start?
Enrol in This Course
₹7,999
₹14,000
Save ₹6,001 · 43% OFF
or ₹667/month · 12-month no-cost EMI
🔒 Secured by Razorpay · 100% refund after 2 sessions if unsatisfied
Everything included
✓ 15 live sessions · 30 hrs total
✓ Lifetime recording access
✓ 4 end-to-end deployment projects
✓ MLOps architecture templates
✓ LinkedIn-verified Platinum badge
✓ Docker + K8s starter kits
✓ Mentor office hours (1hr/week)
✓ Resume & LinkedIn review
✓ Placement referral support
// Alumni Feedback
What Students Say
★★★★★
I knew Docker from work but had never connected it to ML deployments. The FastAPI + Docker + GitHub Actions pipeline from week 3 is now copied verbatim into my company's ML project template. My manager asked who built it. That conversation led directly to a promotion discussion.
SN
Santosh Nambiar
SDE → Senior ML Engineer · Swiggy
★★★★★
Vikram's teaching style is brutally practical — he skips the theory that doesn't matter in production and goes deep on what does. The drift detection week completely changed how I think about model lifecycle. I shipped our company's first model monitoring dashboard the week after the course ended.
LP
Lavanya Pillai
ML Engineer → MLOps Engineer · PhonePe
★★★★★
The Platinum badge on LinkedIn generated more recruiter messages in two weeks than my degree certificate has in two years. "Kubernetes + MLflow + Evidently" on a profile is genuinely rare — most ML people can't deploy their own models. This course makes you the exception.
RJ
Rohit Jain
B.Tech IT → MLOps Engineer · Razorpay
★★★★☆
Week 4 (Kubernetes) is dense and assumes some Linux comfort. I had to rewatch sessions and look up some kubectl commands. But after week 4 clicked, everything in week 5 made complete sense. The monitoring capstone felt like real work — in the best possible way.
GT
Gayatri Trivedi
Data Scientist → AI Platform Engineer · Meesho
// What's Next
Students Also Take