LEECOI
Sign in Register interest
LEECOI
Register interest Sign in
Published 18 weeks

AI Engineer

AI Engineer roles that combine machine learning literacy with LLM application development, RAG systems, vector search, agentic workflows, safety, and production backend engineering.

Register interest All six tracks

Before you start

What this track assumes

Where you learned these does not matter. That you have them does, because the first weeks move quickly.

  • Python, APIs, Git, SQL, JSON, and REST fundamentals
  • Core machine learning concepts and model evaluation basics
  • Comfort reading documentation for fast-moving AI frameworks
  • Ability to build several portfolio applications with deployment-ready code

The roadmap

4 phases across 18 weeks

Each phase ends in a checkpoint: a portfolio artifact, not a quiz.

Phase Focus Checkpoint
01 Weeks 1 to 4 Machine learning and LLM foundations Prompt-evaluated assistant
02 Weeks 5 to 8 Embeddings, vector databases, retrieval Document-grounded chatbot
03 Weeks 9 to 12 LangChain, LangGraph, agents Stateful tool-using agent
04 Weeks 13 to 18 Deployment, evaluation, safety Production AI application

The syllabus

7 modules, in order

Every module names the skills it covers, the methods and tools it uses by name, the project that follows it, and what you hand in.

01

Machine Learning Foundations for AI Engineers

Skills
  • Core supervised, unsupervised and deep learning concepts
  • Model evaluation and responsible AI basics
  • Python engineering, APIs, async programming, testing
  • Cloud notebooks, GPU basics, and cost-aware experimentation
Methods and tools
  • Linear and Logistic Regression
  • Random Forest
  • XGBoost, LightGBM, CatBoost
  • SVM
  • K-Means
  • PCA
  • CNN
  • RNN, LSTM, GRU
  • Transformer
Project
Build a classical machine learning baseline and a deep learning model for the same classification task.
Deliverables
  • Metrics comparison
  • Error analysis
  • Reproducible training scripts

02

LLM Foundations and Prompt Engineering

Skills
  • Tokenization, context windows, sampling, structured outputs
  • Prompt patterns: zero-shot, few-shot, role, task, and context format
  • Evaluation of prompts with test sets and rubrics
  • Safety, privacy, and hallucination control
Methods and tools
  • Transformer
  • Self-Attention
  • Multi-Head Attention
  • Beam Search
  • Top-k Sampling
  • Top-p and Nucleus Sampling
  • Temperature Scaling
Project
Create a prompt-evaluated assistant for a specific business workflow.
Deliverables
  • Prompt library
  • Evaluation cases
  • Quality scorecard
  • Failure analysis

03

Embeddings, Vector Search, and Vector Databases

Skills
  • Text chunking, embedding generation, metadata design
  • Semantic search, hybrid retrieval, reranking
  • Vector database indexing, filtering, namespaces, scaling
  • Data ingestion pipelines and document freshness
  • Pinecone, FAISS, Chroma, Weaviate, Milvus, Qdrant, pgvector, Redis Vector, Elasticsearch and OpenSearch kNN
Methods and tools
  • Approximate Nearest Neighbor
  • HNSW
  • IVF and Inverted File Index
  • Product Quantization
  • Scalar Quantization
  • Cosine Similarity
  • Dot Product Similarity
  • Euclidean and L2 Distance
  • BM25
  • Reciprocal Rank Fusion
  • Maximal Marginal Relevance
Project
Build a semantic search system over PDFs, documents, or internal knowledge-base data.
Deliverables
  • Chunking strategy
  • Embedding pipeline
  • Vector index
  • Search quality evaluation

04

Retrieval-Augmented Generation

Skills
  • RAG architecture: loaders, splitters, embeddings, retrievers, generator
  • Grounded answering with citations and source attribution
  • Hybrid search, query rewriting, multi-query retrieval
  • RAG evaluation: faithfulness, context precision, answer relevance, latency and cost
Methods and tools
  • Naive RAG
  • Advanced RAG
  • Corrective RAG
  • Self-RAG
  • GraphRAG
  • Agentic RAG
  • Hybrid Dense and Sparse Retrieval
  • Cross-Encoder Reranking
  • ColBERT-style late interaction
  • Query Expansion
  • HyDE
Project
Build a production-style RAG chatbot that answers from a document collection with citations.
Deliverables
  • LangChain or LlamaIndex implementation
  • Vector database integration
  • RAG evaluation report
  • Deployment demo

05

LangChain, LangGraph, and Agentic Workflows

Skills
  • LangChain Expression Language, chains, and retrievers
  • Tools, function calling, structured output parsers
  • LangGraph state, nodes, edges, conditional routing, checkpoints
  • Human-in-the-loop, memory, retries, guardrails, observability
Methods and tools
  • ReAct Agent
  • Plan-and-Execute Agent
  • Tool-Calling Agent
  • Reflection Agent
  • Supervisor and Router Agent
  • Multi-Agent Workflow
  • State Machine Graph
  • DAG workflow
  • Tree-of-Thought style search concepts
Project
Create a multi-step research or support agent using LangGraph with tool usage and human approval.
Deliverables
  • Graph diagram
  • State schema
  • Tool definitions
  • Logs and traces
  • Safe fallback behaviour

06

AI Application Backend and Production Deployment

Skills
  • FastAPI or Node backend for AI applications
  • Streaming responses, queues, caching, rate limits
  • Authentication, authorization, secrets management
  • Observability with tracing, cost dashboards, feedback capture
Methods and tools
  • Caching policies
  • Semantic cache
  • Rate limiting algorithms
  • Circuit breaker pattern
  • Canary deployment
  • A/B testing
Project
Deploy a RAG or agentic AI application with authentication, logs, and feedback collection.
Deliverables
  • API service
  • Frontend or demo UI
  • Dockerized deployment
  • Monitoring dashboard

07

AI Safety, Security, and Governance

Skills
  • Prompt injection and data exfiltration defences
  • Handling of personal data, content moderation, access control
  • Model cards, audit logs, evaluation gates
  • Compliance-aware design for enterprise AI
Methods and tools
  • Prompt-injection detection heuristics
  • Redaction methods for personal data
  • Policy classifiers
  • Jailbreak test suites
  • Adversarial evaluation
Project
Add security and governance controls to an AI assistant.
Deliverables
  • Threat model
  • Red-team test cases
  • Guardrails implementation
  • Audit-ready report

What you end up with

3 capstone projects

  • Enterprise RAG Assistant

    A document-grounded AI assistant using embeddings, a vector database, retrieval, reranking, grounded generation, and source citations.

    • Document ingestion pipeline
    • Pinecone, FAISS, or another vector database integration
    • RAG quality evaluation
    • Deployed demo with feedback capture

  • LangGraph Multi-Agent Workflow

    A stateful AI workflow that routes tasks, calls tools, remembers state, handles failures, and uses human approval for risky actions.

    • LangGraph architecture diagram
    • Tool-calling workflow
    • Checkpointing and logging
    • Security and guardrail tests

  • AI Production Readiness Review

    Turn one AI application into a production-ready system with cost controls, observability, prompt-injection defences, and governance documentation.

    • Threat model
    • Cost and latency dashboard
    • Evaluation suite
    • Runbook and model card

Read the full syllabus

Register interest and we will send the complete week-by-week document, then arrange a short call to work out whether this is the right track.

Register interest