ML CI/CD Pipeline: Automate Model Training, Deployment & Monitoring

ML CI/CD pipeline: automate training, deployment, and monitoring

Picture this: you train a new model version, manually upload it to the server, and within an hour users complain about quality degradation. Rolling back means another manual process, losing hours. That's exactly why we build ML CI/CD pipelines that automate training, testing, and deployment, ensuring stability and speed. This approach aligns with the MLOps concept, merging development and operations.

Why CI/CD for ML differs from classical CI/CD

Classical CI/CD tests code. In ML, we also test data, metrics, and inference performance. Models can degrade due to data drift, so the pipeline must trigger retraining on a schedule or when the dataset changes. Each stage has explicit success/failure criteria, and without passing all gates the model never reaches production.

Step-by-step plan for building ML CI/CD

1. Audit current processes — identify manual steps, bottlenecks, and SLAs.
2. Choose the stack — select CI system (GitHub Actions, GitLab CI), orchestrator (Kubeflow, Airflow), and registry (MLflow, DVC).
3. Set up data validation — integrate Great Expectations or similar tools for automatic schema and distribution checks.
4. Automate training — write scripts to run experiments, log metrics, and artifacts.
5. Model evaluation gates — compare the new model against the current production model using thresholds for F1, precision, recall, and latency.
6. Shadow testing — run the new model in parallel on real traffic without affecting users.
7. Canary deployment with auto-rollback — gradually increase traffic with monitoring and automatic rollback on failure.

How we automate training and deployment: a case study

Consider a real project for a retail client: a demand forecasting model was manually updated once a week, often with errors. We deployed a pipeline on GitHub Actions with self-hosted GPU runners. On a push to the train branch, data validation runs (Great Expectations), followed by training with automatic logging to MLflow, then an evaluation gate: F1 must be at least 0.92. If passed, the model is registered and deployed to staging, where integration tests run. On success, a canary (5% of production traffic) is released, monitoring p95 latency and conversion rates. If degradation is detected, the system rolls back automatically within 2 minutes. Result: release time dropped from 4 hours to 15 minutes, incidents decreased by 80%, and infrastructure costs fell by 30% (saving 2 million RUB per year).

Tools we use

Comparison: Kubeflow Pipelines executes steps 1.7x faster than Airflow due to caching and native GPU support.

How is the model tested in CI?

Data validation. Before training, we check schema, feature distributions, and outliers. If data fails validation, the pipeline stops and the team gets an alert.

Model evaluation gates. The new version is compared to the current production model: F1, precision, and recall must not degrade more than 1-2%. If the model is more accurate but latency p95 doubles, it does not pass.

Shadow testing. Production traffic is replayed against the new version in parallel without impacting users. We compare prediction distributions — significant deviations trigger additional review.

Deployment and rollback strategies

Automatic rollback triggers when business metrics (CTR, conversion) drop, inference error rate rises, or latency SLA is exceeded (p99 > 200ms). We guarantee that a bad model will not remain in production longer than 5 minutes.

# Monitoring and auto-rollback
if current_model_metrics['f1'] < production_model_metrics['f1'] * 0.97:
    model_registry.transition_to_stage(current_version, 'Archived')
    model_registry.transition_to_stage(previous_version, 'Production')
    alert_team("Auto-rollback triggered")

What's included in our work

Our service includes: auditing current processes, designing the pipeline, configuring tools, writing configuration and code, documentation, team training, and 2 months of warranty support. Contact us — we'll assess your project and propose a turnkey solution.

Timelines

Basic pipeline (training + staging deployment): from 1 week. Full pipeline with testing, canary, and auto-rollback: from 3 weeks. Enterprise-grade on Kubeflow integrated into CI/CD: from 6 weeks. Get a consultation — we'll refine timelines for your stack.

MLOps: Infrastructure for Training, Deploying, and Monitoring ML Models

The model is trained, metrics — F1 0.94 on validation. Three months later in production, quality drops by 12%. No one knows when — there is no monitoring. It's impossible to retrain quickly — the training script is in a Jupyter notebook of a data scientist who has already left. Data for retraining is collected manually from three disparate systems. About half of the projects come to us with this pain. We build a turnkey MLOps platform: from experiment tracking to automatic deployment and data drift monitoring. We will assess your infrastructure in 1–2 weeks, and in 4–6 weeks you will get a basic MLOps core running in production. Our team has 10+ years of experience in ML infrastructure, over 50 implementations.

How does MLOps infrastructure benefit your ML projects?

Experiment Tracking and Reproducibility

Without tracking, an ML project turns into chaos: it's unclear which checkpoint is better, which hyperparameters were used, which dataset. Reproducing a result a month later is a quest.

Why is experiment tracking the foundation of reproducibility?

MLflow is an open source standard for tracking. It logs parameters, metrics, artifacts (models, graphs), and code. MLflow Model Registry is a centralized model storage with versioning and lifecycle stages (Staging → Production → Archived). Deployment via MLflow Serving or integration with external systems.

Typical initialization in code:

import mlflow

mlflow.set_experiment("fraud-detection-v2")
with mlflow.start_run():
    mlflow.log_params({"learning_rate": 3e-4, "batch_size": 64, "epochs": 10})
    mlflow.log_metric("val_f1", val_f1, step=epoch)
    mlflow.pytorch.log_model(model, "model")

This is the minimum. In production, we add logging of system metrics (GPU utilization, memory), dataset (hash, version), code (git commit hash). Weights & Biases — richer UI, collaboration features, sweep for hyperparameter optimization. MLflow — for on-premise deployment without external dependencies.

DVC (Data Version Control) — versioning of data and models on top of git. Data is stored in S3/GCS/Azure Blob, only metadata (hashes) in git. dvc repro reproduces the entire pipeline from raw data to metrics.

To ensure reproducibility of training, fix random seeds (torch.manual_seed, numpy.random.seed, random.seed) and record them in experiment metadata. Without this, debugging irregular results is painful. Log the dataset version (DVC hash) and git commit — then any experiment can be reproduced down to the byte.

Pipeline Orchestration: Kubeflow, Airflow, Prefect

A pipeline orchestrator becomes necessary when: A 100-line training script in cron is fine for simple tasks. But as soon as you have a multi-step pipeline (data loading → preprocessing → feature engineering → training → validation → deployment if quality above threshold), you need an orchestrator with retry logic, visualization, and alerts.

Kubeflow — Kubernetes-native orchestrator for ML (see Kubeflow). Each step is a Docker container. Supports parallel steps, conditional branches, artifacts between steps. Integrates with Katib (AutoML), KServe (serving), Feast (feature store).

Apache Airflow — more general DAG orchestrator. Wide ecosystem of operators (S3, Spark, DBT, Kubernetes). Easier to deploy if Airflow already exists in the company.

Prefect / Metaflow — less boilerplate. Prefect 2.x with @flow and @task decorators — quick start for small teams.

Typical training pipeline architecture on Kubeflow:

1. Data ingestion component — fetches data from S3/DB, validates schema via Great Expectations
2. Preprocessing component — transformations, normalization, train/val/test split
3. Training component — training on GPU, logging to MLflow
4. Evaluation component — metric calculation, comparison with baseline in Model Registry
5. Conditional deployment — deploy only if new model is better than current by >2% F1

Each component is a separate Docker image. Pipeline is versioned in git. Scheduled run (retraining once a week on new data) or manual.

Model Registry and Lifecycle Management

Model Registry is not just a checkpoint store. It is a centralized system that knows:

• Which model is currently in production (and with what metrics)
• History of all versions with training parameters
• Metadata: dataset, git commit, validation results
• Lifecycle stage: None → Staging → Production → Archived

MLflow Model Registry — standard. For enterprise — Vertex AI Model Registry (GCP), SageMaker Model Registry (AWS), Azure ML Model Registry.

Model promotion through stages: automatically move model to Staging after successful eval, then manual or automatic (during A/B test) promotion to Production. Rollback — switch to previous Production version in seconds.

Serving: From FastAPI to Triton Inference Server

Simple case. FastAPI + PyTorch/ONNX on one server — 80% of production ML deployments are exactly that. Sufficient for most tasks with load up to 100 req/s.

from fastapi import FastAPI
import onnxruntime as ort

app = FastAPI()
session = ort.InferenceSession("model.onnx", providers=["CUDAExecutionProvider"])

@app.post("/predict")
async def predict(request: PredictRequest):
    inputs = preprocess(request.text)
    outputs = session.run(None, {"input_ids": inputs})
    return {"label": postprocess(outputs)}

Triton Inference Server — production standard for high loads (500+ req/s). Dynamic batching, concurrent model execution, model ensemble. Supports TensorRT, ONNX, PyTorch TorchScript, TensorFlow SavedModel.

KServe — Kubernetes-native ML serving with autoscaling, canary deployments, A/B testing out of the box. Scale-to-zero for inactive models — savings on infrastructure up to 40% annually for a project with 10 models.

Monitoring: Data Drift, Model Drift, Infrastructure Metrics

Monitoring — what is usually done last and regretted first. Three levels.

Infrastructure monitoring. Latency (P50/P95/P99), throughput (req/s), error rate (4xx, 5xx), GPU/CPU utilization. Prometheus + Grafana — standard. Alert when P99 latency > threshold or error rate > 1%.

Data drift monitoring. Distribution of input data changes over time. Detect via PSI (Population Stability Index) for numerical features: PSI > 0.2 — strong drift. Chi-squared test for categorical, Kolmogorov-Smirnov test for continuous. Evidently AI — open source library with ready-made drift tests.

Model drift monitoring. If ground truth is delayed (e.g., we know conversion after a week) — monitor real metrics. If not — surrogate metrics: distribution of prediction scores, proportion of confident predictions.

Alerting. Three levels: INFO (minor drift, log it), WARNING (significant, notify team), CRITICAL (quality dropped below threshold — automatic switch to fallback model).

Why is data drift monitoring important?

Without it, you learn about model degradation only from user complaints or ringing SLA. A drift alert allows you to retrain the model in advance, before errors start causing losses. In one of our projects, PSI monitoring detected drift 2 days after a data source change — this saved the campaign.

Feature Store

Feature Store solves the training-serving skew problem. If preprocessing during training and inference is implemented in two different places — divergence is inevitable.

A Feature Store is needed when:

• Several models use the same features
• Features are computed from streaming data (real-time)
• Large team with different people on feature engineering and model training

Feast — open source Feature Store. Offline store (S3 + Parquet) for training, online store (Redis, DynamoDB) for low-latency inference. Feature definitions as code, materialization job syncs offline → online.

Tecton (commercial), Vertex AI Feature Store (GCP), SageMaker Feature Store (AWS) — managed options with less ops overhead.

CI/CD for ML

ML CI/CD is regular CI/CD plus specific ML steps.

ML-specific checks in CI:

• Reproducibility check: run training with a fixed seed, result must match
• Data validation: Great Expectations or Pandera on schema/distribution checks
• Model performance check: automatic eval on holdout, block merge if degradation > threshold
• Latency regression test: inference must meet SLA

GitOps for deployment. Merge to main → CI triggers training → eval → if passes → automatic deployment to Staging → smoke tests → manual promotion to Production or automatic upon successful canary.

Tools: GitHub Actions / GitLab CI for CI, ArgoCD for GitOps deployment on Kubernetes.

What's Included in MLOps Platform Development

We provide a full cycle of work, documentation, and team training.

Total time from audit to full MLOps platform: 3–5 months. Also possible phased launch: basic level (tracking + serving) in 4–6 weeks.

Cost is calculated individually based on data volume, number of models, and infrastructure requirements. Order an MLOps infrastructure audit — get a roadmap in 1–2 weeks. Contact us for a project assessment — we will send a preliminary estimate within 2 business days.

Note: warranty on architectural solutions — 12 months. We provide integration certificates with major cloud providers (AWS, GCP, Azure). During our work, we have not lost a single client after the first implementation — the experience of 50+ successful MLOps projects speaks for itself. Get a consultation on building an MLOps platform today.