Curriculum · AI Application Engineer
Learn how it works underneath. Practise it. Then build with it.
Every concept is taught from first principles, practised in guided exercises, and only then used in one small task on the team project, with a mentor beside you. The whole program is built so that, by the end, you can work confidently in a real codebase.
- Not just calling an LLM API.How the model picks its next word.
- Not just writing useEffect.Why React re-renders, and when it shouldn’t.
- Not just “add an index.”Reading the query plan and knowing why it’s slow.
- Not just Docker commands.What a container actually is.
- 84 teaching days, one concept each
- 14 weeks · full-time · hybrid
- Hyderabad
Every day, the same four steps
ONE CONCEPT / DAY
- 1
Learn
classOne concept from first principles, shown as a step-by-step visual, then live code. - 2
Practise
sandboxGuided exercises on that concept alone, checked instantly. Not the project yet. - 3
Apply
team projectOne small task on the project that uses only today’s concept. - 4
Review
mentor + teamYour mentor and teammates review your change. Fix, then merge.
The project task only ever uses what you learned that day. Nothing you haven't been taught.
Every concept, visualised
You see it run, step by step, before you write it.
Each concept is taught as a step-by-step visual: the code on one side, what the computer is actually doing on the other. Below is a real one from our orientation class: what React does when you tap a filter.
One step at a time
Go forward and back at your own pace, in class and in the recording.
The invisible made visible
Memory, the component tree and the screen change together, so you see cause and effect.
Then watch it break
Each visual ends with a real bug, so you learn to check the output, not just the code.
How support changes over 14 weeks
Lots of help at the start. Less as you get stronger.
In week 1 you're told exactly which files to open, and a mentor walks you through them. By week 14 you plan the work yourself, like an engineer on a real team.
- Week 1: Guided stage, hands-on help at 100%
- Week 2: Guided stage, hands-on help at 96%
- Week 3: Guided stage, hands-on help at 92%
- Week 4: Supported stage, hands-on help at 78%
- Week 5: Supported stage, hands-on help at 74%
- Week 6: Supported stage, hands-on help at 70%
- Week 7: Supported stage, hands-on help at 66%
- Week 8: Supported stage, hands-on help at 62%
- Week 9: Independent stage, hands-on help at 48%
- Week 10: Independent stage, hands-on help at 44%
- Week 11: Independent stage, hands-on help at 40%
- Week 12: Independent stage, hands-on help at 36%
- Week 13: Independent stage, hands-on help at 32%
- Week 14: Independent stage, hands-on help at 28%
Week · bar height = how much hands-on help you get
Weeks 1–3
Guided
- Your task names the exact files to change
- A mentor walks through that code with you first
- You work in pairs
Weeks 4–8
Supported
- Your task names the part of the project; you find the files
- A mentor checks in with you throughout the day
- Pair debugging whenever you’re stuck
Weeks 9–14
Independent
- You plan how to build it
- Your mentor reviews the plan before you start
- You work like an engineer on a real team
ALL 14 WEEKSYou practise a concept before you apply itEvery change is reviewed before it's mergedSunday is a catch-up day
A real day · Week 1, Thursday
What "working on the project" actually means in week 1
LearnMorning
Async JavaScript and errors
A step-by-step visual of the event loop and a promise failing, then live code: async/await, and why an empty catch block hides real problems.
PractiseLate morning
Short exercises
Catch an error, show it to the user, retry once, log it. Each one checks your answer instantly.
ApplyAfternoon
The upload screen fails silently
Your mentor shows you the screen and the two files involved. You make the failure visible to the user.
ReviewEnd of day
Comments, fixes, merge
A mentor and a teammate review your change. You fix their comments and it goes in. That’s the day.
Week by week
14 weeks, one concept at a time
GuidedSupportedIndependent
Engineering foundations
1 · LEARN
- Git from first principles
- Semantic HTML and CSS layout
- Values, copies and closures in JavaScript
- Promises, async/await and the event loop
- HTTP and debugging in the browser
- How code review works
2 · PRACTISE
Git drills, an accessible document card, and short bug hunts: a copy that changes the original, an async loop that doesn’t wait.
3 · APPLY ON THE PROJECT
Get the team project running, then fix small, clearly described bugs in the document list and upload page.
Frontend engineering
1 · LEARN
- How React renders: trigger, render, commit
- Keys and component identity
- State as a snapshot, derived vs stored state
- Forms, validation and error states
- Strict TypeScript and discriminated unions
- Validating data at the boundary
2 · PRACTISE
Predict-what-renders exercises and bug hunts: edits jumping to the wrong row, a count that drifts from the list, a double-click that uploads twice.
3 · APPLY ON THE PROJECT
Typed document list with search, filters and counts, and an upload form with validation.
1 · LEARN
- Effects, cleanup and race conditions
- Server data with TanStack Query
- Saving, rollback and error recovery
- Routing and shareable URLs
- Accessible components and dialogs
- Who owns each piece of state
2 · PRACTISE
Bug hunts: an old search result overwriting a new one, a stale list after delete, a Back button that loses your filters.
3 · APPLY ON THE PROJECT
Document detail pages, shareable filters, login, and keyboard-accessible viewer and delete dialogs.
Backend engineering, databases and storage
1 · LEARN
- Python workflow and core language
- How Python really handles objects
- Parsing CSV and JSON, money and dates
- HTTP and FastAPI with Pydantic
- API design: updates, errors, pagination
- Testing APIs with pytest
2 · PRACTISE
Small scripts and bug hunts: shared records from a default argument, an invoice total parsed wrong, an update that wipes fields.
3 · APPLY ON THE PROJECT
A real FastAPI backend behind the admin app you built. Your screens keep working.
1 · LEARN
- SQL tables, keys and constraints
- Joins, aggregates and transactions
- SQLAlchemy sessions
- Database migrations
- Authentication
- Authorization on every request
2 · PRACTISE
SQL on sample data from two teams, and bug hunts: a join that doubles counts, one user reading another’s documents.
3 · APPLY ON THE PROJECT
Documents saved in Postgres, real login, and every endpoint scoped to the right team.
1 · LEARN
- Clear boundaries in backend code
- All-or-nothing operations
- Race conditions and locking
- Retries and idempotency
- Durable background jobs
- File storage and API security
2 · PRACTISE
Two-session race experiments and bug hunts: a retried upload that creates duplicates, data changing between preview and save.
3 · APPLY ON THE PROJECT
Bulk upload with preview, conflict-safe edits, safe retries, background processing and file storage.
Quality, performance and production
1 · LEARN
- Measuring frontend performance
- Query plans and the N+1 problem
- Logs, metrics and diagnosing incidents
- Backend tests on a real database
- Frontend and end-to-end tests
- Debugging across the full stack
2 · PRACTISE
Profile slow screens, read query plans, and diagnose a supplied incident from logs and metrics.
3 · APPLY ON THE PROJECT
Measured speed-ups on the document list and search, structured logging, and test suites in CI.
1 · LEARN
- Containers
- Configuration and secrets
- CI with required checks
- Deploying safely with migrations
- Health checks, shutdown, backups and restore
2 · PRACTISE
Containerise and deploy small sample services first, then run a restore drill.
3 · APPLY ON THE PROJECT
Your app live in the cloud, with automated checks, a rollback plan and a tested restore.
Data engineering and pipelines
1 · LEARN
- Async Python for AI calls
- Tokens, embeddings, attention and sampling
- Prompt design and versioning
- Pipeline design: stages, retries, idempotency
- OCR and chunking
- Structured extraction and measuring quality
2 · PRACTISE
Run your own experiments with model settings and prompts, and fix a slow call that blocks everything else.
3 · APPLY ON THE PROJECT
Uploaded documents are read, split and their key fields extracted, with an accuracy report.
Search, retrieval and RAG
1 · LEARN
- Keyword search
- Vector search
- Hybrid search
- Permission-aware results
- Reranking
- Measuring search quality
2 · PRACTISE
Compare search methods on a sample set and score them.
3 · APPLY ON THE PROJECT
Search that finds the right passage, not just the right file.
1 · LEARN
- How RAG works
- Grounding and citations
- Streaming answers
- Evaluating answers
- Advanced retrieval techniques
2 · PRACTISE
Build and score small RAG pipelines.
3 · APPLY ON THE PROJECT
Ask a question, get an answer that cites the exact page.
LLMs, AI agents and prompt engineering
1 · LEARN
- The agent loop
- Designing tools
- Prompt and context engineering
- Harness engineering
- Memory
- MCP
- Workflows vs agents
2 · PRACTISE
Write a minimal agent from scratch before using any framework.
3 · APPLY ON THE PROJECT
A document agent with its own tools and an MCP server.
1 · LEARN
- Guardrails and prompt injection
- Human review
- Tracing
- Evaluating agents
- Cost and speed
2 · PRACTISE
Attack and defend a sample agent; write evals for it.
3 · APPLY ON THE PROJECT
A human review queue, traces and cost controls, then a release.
1 · LEARN
- Agent frameworks
- Open-weight models
- Release process
- Your portfolio
2 · PRACTISE
Port one workflow to a framework and compare.
3 · APPLY ON THE PROJECT
Final release, portfolio and a live demo of your team’s platform.
EVERY SATURDAYYou show what you built this week. Then the client sends one small change request on that same code: your first taste of client work, on code you already understand.
Checkpoints
Three points where you're interview-ready
After week 8
Full-stack developer
React and TypeScript frontend, FastAPI and Postgres backend, deployed with CI.
After week 11
GenAI Developer
Document processing, search and cited answers, with evaluation to prove they work.
After week 14
AI application engineer
Agents with tools, guardrails, human review and cost control, shipped and demoed.
What you'll have built
One document platform, built a layer at a time
Each layer appears only after you've learned the concepts behind it.
- Weeks 1–3
Frontend engineering
HTML, CSS, JavaScript, React, TypeScript, state, server data, routing, accessibility - Weeks 4–6
Backend engineering
Python, FastAPI, API design, concurrency, retries, background jobs, security - Weeks 5–6
Databases and storage
Data modelling, SQL, transactions, migrations, authorization, file storage - Weeks 7–8
Quality, performance and production
Testing, profiling, query tuning, observability, Docker, CI/CD, deployment, recovery - Week 9
Data engineering and pipelines
Pipeline design, OCR, chunking, structured extraction, accuracy measurement - Weeks 10–11
Search, retrieval and RAG
Keyword, vector and hybrid search, reranking, grounded answers, evaluation - Weeks 9, 12–14
LLMs, AI agents and prompt engineering
Async model calls, how models work, prompt design, tool calling, harnesses, MCP, guardrails, open models
Learn first. Practise next. Then build.
Every day, for 14 weeks.