Course · Full Stack AI
From an empty repository to an AI product in production.
A 12-week full stack AI course: build, deploy, and operate an AI application end to end — API, retrieval, agents, frontend, infrastructure, and monitoring.
- 12 weeks
- Intermediate
- Project-based
- Deployment
- Live cohort
- Duration
- 12 weeksDuration
- Effort
- 8–10 hours per weekEffort
- Level
- IntermediateLevel
- Modules
- 8Modules
The premise
Most AI courses stop at the notebook. This one starts there and keeps going: the service around the model, the database behind the retrieval, the interface in front of it, the pipeline that ships it, and the dashboards that tell you when it breaks at 2am. Learners finish with one deployed application they can defend in an interview, line by line.
Who it is for
- Software engineers moving into AI
- Backend and full stack developers adding LLM features
- Final-year students targeting AI engineering roles
- Teams standardising how they build AI services
Prerequisites
- Comfortable with Python and Git
- Have written and consumed a REST API before
- Basic SQL and command-line familiarity
Outcomes
What you can do at the end.
Design and ship an AI service with authentication, rate limits, and streaming
Build retrieval over private data with a vector store you can operate
Add an agent layer where it earns its keep, and skip it where it does not
Deploy with CI, containers, and environment separation
Instrument cost, latency, and quality, and debug non-deterministic failures
Syllabus
8 modules, in this order.
The AI application stack
6 hours
- Where the model sits in a real architecture
- Provider APIs, streaming, and failure handling
- Cost and latency budgets before you write code
Backend for AI services
10 hours
- FastAPI service design and async patterns
- Streaming responses and cancellation
- Queues, background jobs, and timeouts
- Authentication, rate limiting, and abuse control
Data and retrieval
12 hours
- Ingestion, chunking, and metadata design
- Embeddings and vector stores in production
- Hybrid search and reranking
- Keeping an index fresh without rebuilding it
Agents and tools
10 hours
- The loop from scratch, then LangGraph
- Tool design, schemas, and error surfaces
- MCP servers for reusable capability
Frontend for AI
8 hours
- Streaming UI and optimistic states
- Citations, sources, and trust affordances
- Handling latency and partial failure in the interface
Evaluation and quality
8 hours
- Golden sets and regression suites
- Retrieval metrics and end-to-end task success
- LLM-as-judge, and where it misleads you
Deployment and infrastructure
10 hours
- Containers, CI, and environment separation
- Secrets, config, and provider failover
- Caching layers and cost controls
Operations
8 hours
- Tracing an agent run end to end
- Cost attribution per feature and per user
- Incident response for non-deterministic systems
Capstone
Ship a document intelligence product: ingestion, hybrid retrieval, an agent that answers with citations, a streaming interface, CI deployment, and a dashboard covering cost, latency, and answer quality.
Tools and stack
- Python
- FastAPI
- Next.js
- PostgreSQL + pgvector
- Qdrant
- LangGraph
- MCP
- Docker
- GitHub Actions
- OpenTelemetry
Delivery
How it can run.
Live cohort
Self-paced
Corporate batch
Corporate batches run against your own stack and swap the capstone for an internal use case. Campus delivery includes syllabus mapping, rubrics, and question banks — and faculty training if you want to own it afterwards.
FAQ
Full Stack AI Engineering — questions
What does full stack AI mean in this course?
Everything between a user's request and a model's response, plus everything that keeps it running: the API service, retrieval and data layer, agent orchestration, the interface, deployment, and observability. It does not mean training foundation models — that is a different discipline with different economics.
Do I need machine learning experience?
No. This is an engineering course. You need to be a competent Python developer; you do not need to have trained a model. The maths that matters here is arithmetic on token costs and latency budgets.
How much does the API usage cost during the course?
Budget roughly 20 to 40 US dollars across twelve weeks. Every lab has a local-model fallback using Ollama for learners who would rather spend nothing, and the cost-control module exists precisely because this question matters in production too.
Is this course available as a corporate batch?
Yes. Corporate batches run against your own stack — your cloud, your vector store, your CI — and the capstone is usually a real internal use case rather than the default brief.
What do I have at the end?
One deployed application, a repository with commit history that shows how it was built, an evaluation suite, and a monitoring dashboard. That combination is far more persuasive in interviews than a folder of notebooks.
Which frameworks are taught?
FastAPI, Next.js, LangGraph, and MCP, with pgvector and Qdrant for retrieval. Frameworks are taught after the underlying mechanism, so when one is replaced next year the knowledge transfers.
Other courses
8 weeks
Generative AI
An 8-week generative AI course covering LLMs, embeddings, RAG, fine-tuning, multimodal models, and safety — with working code at every stage.
8 weeks
Agentic AI
An 8-week agentic AI course on tool use, reasoning loops, memory, multi-agent coordination, MCP, and the evaluation that keeps autonomous systems trustworthy.
6 weeks
Forward Deployed AI Engineer
A 6-week course for forward deployed AI engineers: scoping in the field, prototyping against messy customer data, integration, demos that convince, and handover that lasts.
Run this
for your people.
Tell us the cohort size, the hours available, and whether this is a corporate batch or a campus program. Scope and price come back in writing within two working days.
Request a proposal