conversations listlengths 2 2 | metadata dict | id stringlengths 36 36 |
|---|---|---|
[
{
"from": "human",
"value": "How do I build a production multi-model router that routes requests to different LLMs based on query complexity — e.g., simple queries to a small fast model, complex ones to a large model?"
},
{
"from": "gpt",
"value": "## Multi-Model Router for LLM Serving\n\nA rout... | {
"category": "model_serving",
"context": "building a multi-model router based on request complexity"
} | 73b996ad-57b0-492a-890e-d5805a8e00ae |
[
{
"from": "human",
"value": "How do I run a statistically rigorous A/B test for two ML models in production? I need traffic splitting, metric collection, and significance testing."
},
{
"from": "gpt",
"value": "## A/B Testing ML Models in Production\n\n### Architecture Overview\n\n```\nRequest →... | {
"category": "ab_testing_ml",
"context": "A/B testing ML models in production with statistical significance"
} | 4f1eb1e0-d16e-4012-91fb-d2ee88979170 |
[
{
"from": "human",
"value": "Compare vLLM, Text Generation Inference (TGI), and SGLang for high-throughput LLM serving. What batching strategies do they use?"
},
{
"from": "gpt",
"value": "## High-Throughput LLM Serving: vLLM vs TGI vs SGLang\n\n### The Core Problem\n\nNaive LLM serving is seque... | {
"category": "llm_inference",
"context": "high-throughput LLM serving with vLLM, TGI, and SGLang"
} | 19268553-32d6-4a00-aaef-1ebbaf42de49 |
[
{
"from": "human",
"value": "How do I optimize a PyTorch model for production inference using INT8 quantization, AWQ, and ONNX export? What are the trade-offs?"
},
{
"from": "gpt",
"value": "## ML Model Optimization for Production Inference\n\n### 1. Dynamic INT8 Quantization (CPU)\n\nFastest pa... | {
"category": "model_optimization",
"context": "quantization, pruning, and ONNX export for inference optimization"
} | 88da5cc5-e502-403a-8e3c-6662488e2384 |
[
{
"from": "human",
"value": "How do I set up Triton Inference Server for serving multiple models with dynamic batching and optimal GPU utilization?"
},
{
"from": "gpt",
"value": "## Triton Inference Server: Multi-Model Serving\n\nTriton supports TensorRT, ONNX, PyTorch, TensorFlow, and Python ba... | {
"category": "model_serving",
"context": "Triton Inference Server for multi-model GPU serving"
} | d996732e-0385-4004-9266-4451d49a7cff |
[
{
"from": "human",
"value": "How do I detect data drift and concept drift in production using Evidently? Explain PSI and KL divergence metrics."
},
{
"from": "gpt",
"value": "## Production ML Monitoring: Drift Detection with Evidently\n\n### Installing Evidently\n\n```bash\npip install evidently... | {
"category": "model_monitoring",
"context": "data drift detection and monitoring in production"
} | 64d4f0f7-8915-4508-967a-5f5bc5b5cdeb |
[
{
"from": "human",
"value": "How do I run a statistically rigorous A/B test for two ML models in production? I need traffic splitting, metric collection, and significance testing."
},
{
"from": "gpt",
"value": "## A/B Testing ML Models in Production\n\n### Architecture Overview\n\n```\nRequest →... | {
"category": "ab_testing_ml",
"context": "A/B testing ML models in production with statistical significance"
} | 44b6d12e-8235-43a8-824b-bbadfabad30f |
[
{
"from": "human",
"value": "How do I detect data drift and concept drift in production using Evidently? Explain PSI and KL divergence metrics."
},
{
"from": "gpt",
"value": "## Production ML Monitoring: Drift Detection with Evidently\n\n### Installing Evidently\n\n```bash\npip install evidently... | {
"category": "model_monitoring",
"context": "data drift detection and monitoring in production"
} | 80d257ba-2430-40df-ac10-be6631220509 |
[
{
"from": "human",
"value": "Compare vLLM, Text Generation Inference (TGI), and SGLang for high-throughput LLM serving. What batching strategies do they use?"
},
{
"from": "gpt",
"value": "## High-Throughput LLM Serving: vLLM vs TGI vs SGLang\n\n### The Core Problem\n\nNaive LLM serving is seque... | {
"category": "llm_inference",
"context": "high-throughput LLM serving with vLLM, TGI, and SGLang"
} | e8731f5a-379f-4d4b-a79d-2b04e5afbcc2 |
[
{
"from": "human",
"value": "How do I detect data drift and concept drift in production using Evidently? Explain PSI and KL divergence metrics."
},
{
"from": "gpt",
"value": "## Production ML Monitoring: Drift Detection with Evidently\n\n### Installing Evidently\n\n```bash\npip install evidently... | {
"category": "model_monitoring",
"context": "data drift detection and monitoring in production"
} | 618546c8-b6f7-4583-b88a-f37545de72b6 |
[
{
"from": "human",
"value": "How do I deploy a scikit-learn model on Kubernetes using KServe? Include auto-scaling and canary rollout configuration."
},
{
"from": "gpt",
"value": "## Deploying ML Models on Kubernetes with KServe\n\n### Prerequisites\n\n```bash\n# Install KServe (after cert-manag... | {
"category": "kubernetes_ml",
"context": "deploying ML models on Kubernetes with KServe and auto-scaling"
} | 79b2b4c9-41a9-4d71-9d63-740c9c72c8ed |
[
{
"from": "human",
"value": "Compare vLLM, Text Generation Inference (TGI), and SGLang for high-throughput LLM serving. What batching strategies do they use?"
},
{
"from": "gpt",
"value": "## High-Throughput LLM Serving: vLLM vs TGI vs SGLang\n\n### The Core Problem\n\nNaive LLM serving is seque... | {
"category": "llm_inference",
"context": "high-throughput LLM serving with vLLM, TGI, and SGLang"
} | 54e671a4-87df-49c0-847f-a6df66161aca |
[
{
"from": "human",
"value": "How do I build a CI/CD pipeline for ML models using GitHub Actions and DVC? I need model validation gates before deploying to production."
},
{
"from": "gpt",
"value": "## CI/CD for ML: GitHub Actions + DVC + Validation Gates\n\n### Repository Structure\n\n```\nml-pr... | {
"category": "cicd_ml",
"context": "CI/CD pipelines for ML with GitHub Actions, DVC, and model validation"
} | 8a35693b-9532-4462-a19e-1fdbffabf7e4 |
[
{
"from": "human",
"value": "How do I build a CI/CD pipeline for ML models using GitHub Actions and DVC? I need model validation gates before deploying to production."
},
{
"from": "gpt",
"value": "## CI/CD for ML: GitHub Actions + DVC + Validation Gates\n\n### Repository Structure\n\n```\nml-pr... | {
"category": "cicd_ml",
"context": "CI/CD pipelines for ML with GitHub Actions, DVC, and model validation"
} | 38d03286-4ae7-4b12-954a-db760027fad4 |
[
{
"from": "human",
"value": "How do I set up Triton Inference Server for serving multiple models with dynamic batching and optimal GPU utilization?"
},
{
"from": "gpt",
"value": "## Triton Inference Server: Multi-Model Serving\n\nTriton supports TensorRT, ONNX, PyTorch, TensorFlow, and Python ba... | {
"category": "model_serving",
"context": "Triton Inference Server for multi-model GPU serving"
} | 09f182a6-4c59-4a55-8dc4-32408831914d |
[
{
"from": "human",
"value": "Compare Feast, Tecton, and Hopsworks as feature stores. What is point-in-time correctness and why does it matter for ML training?"
},
{
"from": "gpt",
"value": "## Feature Store Comparison: Feast vs Tecton vs Hopsworks\n\n### Why Feature Stores Exist\n\nWithout a fea... | {
"category": "feature_stores",
"context": "feature store selection and point-in-time correctness"
} | c729461f-3e68-4c81-a685-b7cb34d70b9a |
[
{
"from": "human",
"value": "How do I run a statistically rigorous A/B test for two ML models in production? I need traffic splitting, metric collection, and significance testing."
},
{
"from": "gpt",
"value": "## A/B Testing ML Models in Production\n\n### Architecture Overview\n\n```\nRequest →... | {
"category": "ab_testing_ml",
"context": "A/B testing ML models in production with statistical significance"
} | 51f92ec6-1d3c-492d-8016-580c669f60b2 |
[
{
"from": "human",
"value": "Walk me through setting up MLflow for experiment tracking: logging parameters, metrics, artifacts, and promoting models through the model registry."
},
{
"from": "gpt",
"value": "## MLflow Experiment Tracking and Model Registry\n\n### Setup\n\n```bash\npip install ml... | {
"category": "mlflow_tracking",
"context": "MLflow experiment tracking and model registry"
} | 37141001-8397-421b-a2f6-ee11409099eb |
[
{
"from": "human",
"value": "How do I optimize ML inference costs in production? Cover GPU utilization, spot instances, model caching, and request batching strategies."
},
{
"from": "gpt",
"value": "## ML Inference Cost Optimization\n\n### 1. GPU Utilization — Fix the #1 Cost Leak\n\nMost teams ... | {
"category": "cost_optimization",
"context": "reducing ML inference costs with GPU optimization and batching"
} | 49ab9792-4df1-4fd4-96b0-c0dc86538886 |
[
{
"from": "human",
"value": "How do I implement observability for ML inference services using OpenTelemetry? I need latency tracking, P99 monitoring, and alerting."
},
{
"from": "gpt",
"value": "## ML Observability with OpenTelemetry\n\n### Setup\n\n```bash\npip install opentelemetry-api opentel... | {
"category": "observability",
"context": "OpenTelemetry tracing and P99 monitoring for ML services"
} | d5391517-03b8-478e-b106-4249ae05ee8b |
[
{
"from": "human",
"value": "Compare vLLM, Text Generation Inference (TGI), and SGLang for high-throughput LLM serving. What batching strategies do they use?"
},
{
"from": "gpt",
"value": "## High-Throughput LLM Serving: vLLM vs TGI vs SGLang\n\n### The Core Problem\n\nNaive LLM serving is seque... | {
"category": "llm_inference",
"context": "high-throughput LLM serving with vLLM, TGI, and SGLang"
} | 4515f723-1b5c-48b9-b025-d015eb59a533 |
[
{
"from": "human",
"value": "Compare vLLM, Text Generation Inference (TGI), and SGLang for high-throughput LLM serving. What batching strategies do they use?"
},
{
"from": "gpt",
"value": "## High-Throughput LLM Serving: vLLM vs TGI vs SGLang\n\n### The Core Problem\n\nNaive LLM serving is seque... | {
"category": "llm_inference",
"context": "high-throughput LLM serving with vLLM, TGI, and SGLang"
} | 2d95eeae-004f-4fb0-b415-ea3ce205dc35 |
[
{
"from": "human",
"value": "How do I implement a progressive rollout for an ML model with automated rollback triggered by metric degradation? I want to go from 5% to 100% traffic automatically."
},
{
"from": "gpt",
"value": "## Progressive Rollout with Automated Rollback\n\n### Rollout Controll... | {
"category": "deployment_patterns",
"context": "progressive rollout with automated rollback on metric degradation"
} | 37d9dbe6-3d5a-4d58-bb17-2e753e2eacf7 |
[
{
"from": "human",
"value": "How do I optimize ML inference costs in production? Cover GPU utilization, spot instances, model caching, and request batching strategies."
},
{
"from": "gpt",
"value": "## ML Inference Cost Optimization\n\n### 1. GPU Utilization — Fix the #1 Cost Leak\n\nMost teams ... | {
"category": "cost_optimization",
"context": "reducing ML inference costs with GPU optimization and batching"
} | 3e7b1648-d3fe-482e-a26b-f9152a19db41 |
[
{
"from": "human",
"value": "How do I implement a progressive rollout for an ML model with automated rollback triggered by metric degradation? I want to go from 5% to 100% traffic automatically."
},
{
"from": "gpt",
"value": "## Progressive Rollout with Automated Rollback\n\n### Rollout Controll... | {
"category": "deployment_patterns",
"context": "progressive rollout with automated rollback on metric degradation"
} | c45704a7-eb22-440b-8681-096e514ba053 |
[
{
"from": "human",
"value": "How do I implement a progressive rollout for an ML model with automated rollback triggered by metric degradation? I want to go from 5% to 100% traffic automatically."
},
{
"from": "gpt",
"value": "## Progressive Rollout with Automated Rollback\n\n### Rollout Controll... | {
"category": "deployment_patterns",
"context": "progressive rollout with automated rollback on metric degradation"
} | a1d51046-f560-4f42-9de7-2c0dfd454e6f |
[
{
"from": "human",
"value": "How do I set up Triton Inference Server for serving multiple models with dynamic batching and optimal GPU utilization?"
},
{
"from": "gpt",
"value": "## Triton Inference Server: Multi-Model Serving\n\nTriton supports TensorRT, ONNX, PyTorch, TensorFlow, and Python ba... | {
"category": "model_serving",
"context": "Triton Inference Server for multi-model GPU serving"
} | a719a6f3-4701-41f4-96e5-10c3f0feb89d |
[
{
"from": "human",
"value": "I need to serve a fine-tuned LLM in production. What are the trade-offs between FastAPI, BentoML, TorchServe, and vLLM? Which should I use?"
},
{
"from": "gpt",
"value": "## Choosing an LLM Serving Framework\n\nThe right choice depends on your latency, throughput, an... | {
"category": "model_serving",
"context": "choosing a serving framework for ML models"
} | e2f16550-4777-43df-a23f-fbd09d2dbe9c |
[
{
"from": "human",
"value": "How do I detect data drift and concept drift in production using Evidently? Explain PSI and KL divergence metrics."
},
{
"from": "gpt",
"value": "## Production ML Monitoring: Drift Detection with Evidently\n\n### Installing Evidently\n\n```bash\npip install evidently... | {
"category": "model_monitoring",
"context": "data drift detection and monitoring in production"
} | cca64f85-0733-4c45-90a8-899ce1324aa9 |
[
{
"from": "human",
"value": "Compare Feast, Tecton, and Hopsworks as feature stores. What is point-in-time correctness and why does it matter for ML training?"
},
{
"from": "gpt",
"value": "## Feature Store Comparison: Feast vs Tecton vs Hopsworks\n\n### Why Feature Stores Exist\n\nWithout a fea... | {
"category": "feature_stores",
"context": "feature store selection and point-in-time correctness"
} | ff62f51d-394b-482d-b2f8-04ad43275b2f |
[
{
"from": "human",
"value": "How do I build a production multi-model router that routes requests to different LLMs based on query complexity — e.g., simple queries to a small fast model, complex ones to a large model?"
},
{
"from": "gpt",
"value": "## Multi-Model Router for LLM Serving\n\nA rout... | {
"category": "model_serving",
"context": "building a multi-model router based on request complexity"
} | f8ed696c-106f-468f-bdd7-3a8eabd8b6b2 |
[
{
"from": "human",
"value": "How do I build a CI/CD pipeline for ML models using GitHub Actions and DVC? I need model validation gates before deploying to production."
},
{
"from": "gpt",
"value": "## CI/CD for ML: GitHub Actions + DVC + Validation Gates\n\n### Repository Structure\n\n```\nml-pr... | {
"category": "cicd_ml",
"context": "CI/CD pipelines for ML with GitHub Actions, DVC, and model validation"
} | 5e07ac6a-9277-402d-8ac5-7e900ca987ff |
[
{
"from": "human",
"value": "How do I detect data drift and concept drift in production using Evidently? Explain PSI and KL divergence metrics."
},
{
"from": "gpt",
"value": "## Production ML Monitoring: Drift Detection with Evidently\n\n### Installing Evidently\n\n```bash\npip install evidently... | {
"category": "model_monitoring",
"context": "data drift detection and monitoring in production"
} | b9baa4a1-5848-49ae-8990-5dca594d8f0d |
[
{
"from": "human",
"value": "How do I build a production multi-model router that routes requests to different LLMs based on query complexity — e.g., simple queries to a small fast model, complex ones to a large model?"
},
{
"from": "gpt",
"value": "## Multi-Model Router for LLM Serving\n\nA rout... | {
"category": "model_serving",
"context": "building a multi-model router based on request complexity"
} | 7ae67548-868d-464d-904e-ac25cb256c1b |
[
{
"from": "human",
"value": "Compare vLLM, Text Generation Inference (TGI), and SGLang for high-throughput LLM serving. What batching strategies do they use?"
},
{
"from": "gpt",
"value": "## High-Throughput LLM Serving: vLLM vs TGI vs SGLang\n\n### The Core Problem\n\nNaive LLM serving is seque... | {
"category": "llm_inference",
"context": "high-throughput LLM serving with vLLM, TGI, and SGLang"
} | b8ccafa7-243b-4f18-a897-d6fd71310e6c |
[
{
"from": "human",
"value": "How do I build a CI/CD pipeline for ML models using GitHub Actions and DVC? I need model validation gates before deploying to production."
},
{
"from": "gpt",
"value": "## CI/CD for ML: GitHub Actions + DVC + Validation Gates\n\n### Repository Structure\n\n```\nml-pr... | {
"category": "cicd_ml",
"context": "CI/CD pipelines for ML with GitHub Actions, DVC, and model validation"
} | 904eb744-3f2a-4cb3-aa28-73a72e8ae4cb |
[
{
"from": "human",
"value": "Compare Feast, Tecton, and Hopsworks as feature stores. What is point-in-time correctness and why does it matter for ML training?"
},
{
"from": "gpt",
"value": "## Feature Store Comparison: Feast vs Tecton vs Hopsworks\n\n### Why Feature Stores Exist\n\nWithout a fea... | {
"category": "feature_stores",
"context": "feature store selection and point-in-time correctness"
} | 6ff90240-9d6a-489c-8c33-bf0d02cd2295 |
[
{
"from": "human",
"value": "How do I implement observability for ML inference services using OpenTelemetry? I need latency tracking, P99 monitoring, and alerting."
},
{
"from": "gpt",
"value": "## ML Observability with OpenTelemetry\n\n### Setup\n\n```bash\npip install opentelemetry-api opentel... | {
"category": "observability",
"context": "OpenTelemetry tracing and P99 monitoring for ML services"
} | a5413b66-bedf-453e-baf2-e6e1c00b9523 |
[
{
"from": "human",
"value": "How do I implement a progressive rollout for an ML model with automated rollback triggered by metric degradation? I want to go from 5% to 100% traffic automatically."
},
{
"from": "gpt",
"value": "## Progressive Rollout with Automated Rollback\n\n### Rollout Controll... | {
"category": "deployment_patterns",
"context": "progressive rollout with automated rollback on metric degradation"
} | fe54ab2c-cde0-453f-89a0-4ac98860e1a9 |
[
{
"from": "human",
"value": "How do I deploy a scikit-learn model on Kubernetes using KServe? Include auto-scaling and canary rollout configuration."
},
{
"from": "gpt",
"value": "## Deploying ML Models on Kubernetes with KServe\n\n### Prerequisites\n\n```bash\n# Install KServe (after cert-manag... | {
"category": "kubernetes_ml",
"context": "deploying ML models on Kubernetes with KServe and auto-scaling"
} | ba343a37-d3e7-4aaa-9818-ae894082ec0b |
[
{
"from": "human",
"value": "How do I set up Triton Inference Server for serving multiple models with dynamic batching and optimal GPU utilization?"
},
{
"from": "gpt",
"value": "## Triton Inference Server: Multi-Model Serving\n\nTriton supports TensorRT, ONNX, PyTorch, TensorFlow, and Python ba... | {
"category": "model_serving",
"context": "Triton Inference Server for multi-model GPU serving"
} | 6bdf3446-8970-474c-ac0b-37e439034baa |
[
{
"from": "human",
"value": "Explain blue-green, canary, and shadow deployments for ML models. When should I use each, and how do I implement them?"
},
{
"from": "gpt",
"value": "## ML Model Deployment Patterns\n\n### Blue-Green Deployment\n\nTwo identical environments — only one serves live tra... | {
"category": "deployment_patterns",
"context": "safe deployment strategies for ML models"
} | ed0dacff-04d1-4ee6-a7b7-3e77d3182131 |
[
{
"from": "human",
"value": "How do I run a statistically rigorous A/B test for two ML models in production? I need traffic splitting, metric collection, and significance testing."
},
{
"from": "gpt",
"value": "## A/B Testing ML Models in Production\n\n### Architecture Overview\n\n```\nRequest →... | {
"category": "ab_testing_ml",
"context": "A/B testing ML models in production with statistical significance"
} | bf069ffa-f9de-4586-819c-506e735ab7f4 |
[
{
"from": "human",
"value": "How do I run a statistically rigorous A/B test for two ML models in production? I need traffic splitting, metric collection, and significance testing."
},
{
"from": "gpt",
"value": "## A/B Testing ML Models in Production\n\n### Architecture Overview\n\n```\nRequest →... | {
"category": "ab_testing_ml",
"context": "A/B testing ML models in production with statistical significance"
} | 10b1bff7-3ec5-4c77-8aab-cdbd37870b51 |
[
{
"from": "human",
"value": "How do I optimize a PyTorch model for production inference using INT8 quantization, AWQ, and ONNX export? What are the trade-offs?"
},
{
"from": "gpt",
"value": "## ML Model Optimization for Production Inference\n\n### 1. Dynamic INT8 Quantization (CPU)\n\nFastest pa... | {
"category": "model_optimization",
"context": "quantization, pruning, and ONNX export for inference optimization"
} | 5a8e72c0-c32c-4a71-91e3-c744e38dcf11 |
End of preview. Expand in Data Studio
MLOps Deployment SFT 100K
A synthetic supervised fine-tuning dataset of 100,000 high-quality conversations covering MLOps and ML model deployment — from model serving and inference optimization to monitoring, CI/CD, and production scaling. Designed to train AI assistants that can help ML engineers deploy and operate models at scale.
Dataset Description
This dataset covers the full MLOps lifecycle across 13 specialized categories. Each record follows the ShareGPT format with a practitioner-level question and a detailed response with working code examples.
Categories
| Category | Description |
|---|---|
model_serving |
FastAPI, BentoML, Triton, vLLM, multi-model routing |
mlflow_tracking |
Experiment tracking, model registry, artifact management |
deployment_patterns |
Blue-green, canary, shadow deployments, progressive rollout |
feature_stores |
Feast vs Tecton vs Hopsworks, online/offline features |
model_monitoring |
Data drift detection, Evidently, PSI, concept drift |
kubernetes_ml |
KServe, Seldon Core, auto-scaling for ML workloads |
cicd_ml |
GitHub Actions + DVC, model validation gates |
model_optimization |
Quantization (INT8, AWQ), pruning, ONNX export |
ab_testing_ml |
Traffic splitting, statistical significance for ML |
llm_inference |
vLLM, TGI, SGLang — high-throughput LLM serving |
cost_optimization |
GPU utilization, spot instances, request batching |
observability |
OpenTelemetry for ML, P99 monitoring, alerting |
Format
ShareGPT format:
{
"conversations": [
{"from": "human", "value": "...MLOps question..."},
{"from": "gpt", "value": "...production-ready response with code..."}
],
"metadata": {"category": "...", "context": "..."},
"id": "uuid"
}
Use Cases
- Fine-tuning AI assistants for ML deployment tasks
- Training models to reason about production ML operations
- Building AI-assisted MLOps tooling
- Educating teams on model serving and monitoring best practices
Quality Notes
All responses include working Python, YAML, and bash code using FastAPI, MLflow, vLLM, Kubernetes, Evidently, and other production MLOps tools.
- Downloads last month
- 47