Model serving on the platform and the latency it gives - SkilBrill

Skip to content

Home
Courses
About
Contact
Blog
- Freshers

Search for:

Home - Lessons - Databricks Training - M6 — MLflow, AI and Vector Search - Model serving on the platform and the latency it gives

Databricks Training Model serving on the platform and the latency it gives

Databricks Training

0% Completed

M1 — The Lakehouse and the Workspace

What a lakehouse is and why Delta sits under everything

Workspaces, clusters, pools and what actually gets billed

Interactive notebooks versus jobs and their compute

Unity Catalog as the metadata and permission layer

Navigating the workspace: catalog explorer, jobs, workloads

Where Databricks sits against a warehouse and a plain object store

M2 — Delta Lake in Depth

The transaction log: how Delta gives atomic writes and time travel

Schema enforcement, evolution and the failures they prevent

Optimise, V-Order and Z-ordering for scan performance

Compaction, small file problems and vacuuming safely

Batch versus structured streaming into a Delta table

Handling late data and exactly-once semantics in a stream

M3 — Unity Catalog and Governance

Three-level namespacing: catalog, schema and table

Grants, securable objects and the inheritance model

Row-level filters and column masking

External locations, storage credentials and managed versus external

Lineage: tracing a column back to its source

Auditing access and answering who queried what

M4 — Notebooks, SQL and Modelling

Notebooks as code: cells, magics, parameters and reuse

Widgets and passing parameters from a job

Databricks SQL, dashboards and alerts

Materialised views and the SQL performance model

Modelling: facts, dimensions and slowly changing dimensions

Version control: notebooks against a Git repository

M5 — Jobs, Pipelines and Orchestration

Job tasks, dependencies and multi-task jobs

Schedules, triggers and backfills that will not corrupt state

Retries, timeout policy and alerting on failure

Delta Live Tables: declarative pipelines and expectations

Data quality expectations that reject bad records on purpose

Promoting a job across environments without drift

M6 — MLflow, AI and Vector Search

MLflow tracking: experiments, runs, parameters and metrics

Logging artifacts and registering a model with its signature

Model serving on the platform and the latency it gives

Foundation model endpoints and where they fit in an application

Vector search indexes and incremental sync for a RAG workload

Combining Delta, MLflow and vector search for an AI feature

M7 — Applied Skills

Reading a Spark UI: stages, tasks, skew and shuffle cost

Fixing a job that got slow after the data grew

Choosing partitioning and clustering for a table

Cost control: cluster policies, autoscaling and spot usage

Migrating an existing pipeline to Delta with minimal downtime

Reviewing someone else's notebook for production readiness

M8 — Capstone

Ingest raw files into Delta with schema enforcement and time travel

Model the result in SQL with a fact table and dimension tables

Govern it in Unity Catalog with grants, row filters and lineage

Build a multi-task job with expectations, retries and alerting

Add an MLflow-tracked model and a vector search index

Deliver the runbook, cost estimate and access review

Announcements Course Info

Global Trusted Software Training Partner Since 2022

SkilBrill Training Institute delivers industry-focused technology education that bridges the gap between academic learning and real-world demands. Our structured programs combine foundational theory with practical application, ensuring every graduate is confident and job-ready from day one.

Learn More About Us

© 2026 SkilBrill Training Institute. All Rights Reserved. Privacy Policy Terms Refunds Outcomes

📞 Enroll Now

Start Your Journey

Fill in your details and our team will get back to you within 24 hours

Please enable JavaScript in your browser to complete this form.

Full Name *

Email *

Course Interested In *

Mode of Course *