Artificial Intelligence
Applied AI Engineering — Campus Program
One semester. Build, fine-tune, and defend real AI systems — for university credit.
About this course
A 24-week, 200-hour Campus Program delivered with your university, and credit-bearing: your institution decides the credit quantum. Weave teaches 160 hours online and 24×7; IIT professors and industry experts teach the live 20%, with two on-campus classes. Fourteen modules from foundations through RAG, agents, transformers, fine-tuning, and reasoning, ending in a capstone defended before a panel. Enrolment is through your institution.
What you'll learn
- Design, evaluate, adapt, and deploy production AI systems
- Ship two major projects and two live deployments
- Run a fine-tune and defend the verdict against a measured baseline
- Defend a capstone before IIT faculty, an industry expert, and an external judge
Course content
24 weeks · 200 hours · 153 lessons · 63 videos · 23 guided project activities · 13 quizzes · 23 projects
AI Foundations and Responsible AI
What a language model actually does12 min
The model landscape: hosted, open, and the trade14 min
Break three models on purpose60 min
Cost and latency benchmark60 min
Your responsible-use charter60 min
Foundations check34 min
Open and hosted models side by side150 min
Responsible AI casework150 min
Written responsible-AI standard180 min
Prompt Engineering and Reliable AI Systems
From prompt to prompt system14 min
Structured output that parses every time14 min
Prompt injection and untrusted input12 min
Context engineering: budgeting the window18 min
Build a schema-validated extractor90 min
Write the test suite90 min
Mini-build 1: reliable prompt system120 min
Prompt systems check20 min
Prompt-injection red team150 min
Privacy and data-handling casework140 min
Hold a context budget under pressure110 min
Second prompt system, different task class182 min
Python and Model APIs
Project layout and dependency discipline14 min
Calling model APIs safely16 min
Streaming, async, and timeouts14 min
Typed responses and boundary validation12 min
Build a typed model client from scratch80 min
Add streaming and a timeout budget80 min
Secrets audit on your own repository90 min
Mini-build 2: validated streaming model client150 min
Python and APIs check24 min
Packaging, testing, and CI for AI services18 min
Build the CI pipeline150 min
Second service on a different provider312 min
AI Product and Multi-Step Workflows
The failure taxonomy14 min
Retries done right14 min
Observability for AI systems12 min
Wrap your client: retries, breaker, cost ledger80 min
Chaos lab70 min
Mini-build 3: resilient AI service150 min
Production wrapper check20 min
Tool calling: the model as router14 min
Multi-step workflows that resume14 min
Deterministic versus model-decided control flow12 min
Tracing a multi-step run18 min
Give your service three real tools80 min
Build a resumable pipeline70 min
Tool-using workflow150 min
Tools and workflows check20 min
Incident write-up from an induced outage, traced end to end102 min
Retrieval-Augmented Generation
Why RAG — and when it is the wrong answer16 min
Chunking without destroying meaning16 min
Embeddings and vector search14 min
Citation and refusal14 min
Compare three chunking strategies80 min
Retrieval and generation with mandatory citations80 min
Force a refusal70 min
Major project, part 1: grounded knowledge assistant170 min
RAG foundations check20 min
Parsing PDFs, tables, and code18 min
Multi-corpus ingestion160 min
Retrieval over Indian-language and mixed-script documents120 min
Major project 1: deployed knowledge system302 min
Advanced Retrieval and Knowledge Assistants
Your retrieval is worse than you think14 min
Reranking, filtering, and hybrid search14 min
Query rewriting and multi-hop questions12 min
Build a 50-question evaluation set80 min
Add a reranker and measure80 min
Retrieval evaluation report140 min
Advanced retrieval check20 min
Retrieval over code and structured documents16 min
Your corpus is an attack surface16 min
Hybrid search and multi-hop evaluation150 min
Poison your own corpus, then defend it110 min
Retrieval hardening report188 min
Coding Agents and Autonomous Workflows
The agent loop, honestly16 min
Stop conditions and budgets14 min
What an agent must never do alone12 min
Build a coding agent under test80 min
Make it fail safely78 min
Supervised agent workflow140 min
Agents check20 min
Sandboxing and destructive-action prevention16 min
MCP: the standard tool interface18 min
Memory and context on long-running work16 min
Run your agent on a real repository under CI130 min
Expose your tools as an MCP server100 min
Take an agent past a hundred steps90 min
Agent safety pack110 min
Transformers and Foundation Models
From n-grams to attention18 min
Attention mechanics, worked by hand18 min
Tokenisation and the strange behaviour it explains16 min
Sampling: temperature, top-p, and determinism16 min
Pre-training, scaling, and what open really means16 min
Implement attention from scratch120 min
Tokeniser lab: find where your own language breaks120 min
Run an open model locally120 min
Sampling sweep100 min
Model literacy report360 min
Foundation models check56 min
AI Evaluation and Performance Measurement
Evaluation you can defend16 min
LLM-as-judge: when it is legitimate14 min
Regression testing in CI12 min
Build and label your golden set80 min
Error analysis by hand, before any harness90 min
Wire the harness into CI70 min
Evaluation harness and report88 min
Evaluation check20 min
Public benchmarks and what they do not tell you16 min
Calibrate your judge against human labels140 min
Major project 2 evaluation report174 min
Fine-Tuning and Model Adaptation
When to fine-tune, and when not to18 min
Dataset construction18 min
LoRA and parameter-efficient adaptation16 min
Reading a training run16 min
Build your fine-tuning dataset130 min
Run a LoRA fine-tune130 min
Evaluate against your Module 9 baseline120 min
Adaptation report352 min
Fine-tuning check40 min
Reasoning and Advanced Model Training
Preference tuning: RLHF and DPO18 min
Verifiable rewards (RLVR)18 min
GRPO and group-relative methods16 min
Test-time thinking and reasoning budgets16 min
Build a verifiable-reward task120 min
Preference-pair lab110 min
Reasoning-budget sweep110 min
Reasoning strategy memo280 min
Reasoning check32 min
Generative and Multimodal AI
Diffusion in working terms16 min
Multimodal input16 min
Generation inside a product14 min
Build a document-understanding pipeline110 min
Add a generative component to your capstone100 min
Multimodal component, integrated320 min
Generative and multimodal check24 min
Inference Optimisation and Deployment
Deployment paths for AI apps16 min
Monitoring in production14 min
The README a hiring manager will read12 min
Deploy, break, and fix70 min
Quantisation and serving trade-offs18 min
Drift detection and model cards16 min
Cost per task, and how to halve it18 min
Build and hold a latency budget140 min
Wire monitoring and drift detection120 min
Halve your cost per task without losing quality110 min
Model card and operations runbook186 min
Capstone: Build, Fine-Tune and Defend
What a defensible capstone looks like18 min
Error analysis16 min
The safety section every capstone must contain14 min
Run the error analysis180 min
Defence rehearsal with an adversarial panel150 min
Capstone build600 min
Defence pack and live defence222 min