Learning Paths
A path is a workspace for one goal — a job, a cert, an activity — that pulls together everything you need to train for it: flashcard decks, hands-on labs, coding drills, practice exams, and a curated list of the best books, videos, and courses. One place, so nothing gets lost.
Learning & Cognition
Technology & Computing
AI Hardware — Rack, Interconnect and Identification
The physical layer of an AI datacenter, taught the way the work is done: identify the thing in order to install, verify or replace it. Safety first, then rack, node, the PCIe/SXM/OAM split, and the cabling and testing that are the largest part of the job. Weighted against NVIDIA's NCP-ARI blueprint — the technician cert NVIDIA sells but publishes no training for.
PostgreSQL & pgvector
The database first, vector search second, and deliberately one path. Architecture and the write-ahead log, MVCC and the vacuum/bloat consequences, indexes and reading real query plans — then embeddings, exact search as ground truth, the recall trade in approximate indexes, the filtered-search trap, hybrid retrieval and evaluation.
Godot Game Development
Built around the one idea the engine repeats at every scale — everything is a node, a tree of nodes is a scene, a scene instances inside another. GDScript, the frame loop, signals and decoupling, collision layers and masks (the most common silent failure), animation, UI, performance, and actually exporting a finished build.
AI Infrastructure & Platform Engineer — LLMOps + GPU/Cloud Ops
The combined role in one place — a hands-on GPU/LLMOps lab track (ten runnable runbooks: serverless vLLM, the vLLM core build, SGLang, DCGM monitoring, MIG, NIM/Triton, vLLM on Kubernetes, tuning), the concept decks behind the vocabulary, and the cert paths that feed it. Free-tier-first: most labs run on a 16GB card at $0.
Certified Kubernetes Administrator (CKA)
Pass the hands-on Kubernetes admin cert — Learn + Break-fix labs (the real practice), domain-weighted decks, a kubectl-speed drill, a recall quiz, and free-first resources verified against the v1.35 curriculum.
LFCS — Linux Foundation Certified System Administrator
Pass the hands-on, performance-based Linux sysadmin exam — five domain-weighted decks (106 cards), an 85-question recall exam, hands-on Linux & Bash labs, a study routine, and free-first resources verified against the official Linux Foundation domains.
HashiCorp Terraform Associate (003)
Pass the industry-standard infrastructure-as-code cert — eight objective-aligned decks (258 cards), an 83-question practice exam, the browser-graded HCL drills for writing real config, a study routine, and free-first resources from HashiCorp. Prep at $0 with the local/null/random providers.
Ansible Automation
Master the agentless automation and config-management tool — eight topic decks (208 cards) across fundamentals, inventory, playbooks, variables & Jinja2, roles & collections, modules, control flow, and vault, a topic-weighted practice exam, and the browser-graded YAML drills for writing real playbooks. Learn it at $0; optional cert goal is the Red Hat RHCE (EX294).
PCAP — Certified Associate in Python Programming
Pass the Python Institute PCAP-31-03 — blueprint-weighted decks, a full practice exam, and the live in-browser Python coding track for the hands-on reps, verified against the official OpenEDG syllabus.
CompTIA Security+ (SY0-701)
Pass the foundational cybersecurity cert — domain-weighted decks, a high-yield acronym drill, a full practice exam, a study routine, and free-first resources (Professor Messer + the official objectives).
NVIDIA AI Infrastructure & Operations
The datacenter GPU / AI-infra & operations skill set — cert roadmap (NCA-AIIO → CKA → NCP-AIO), decks, hands-on labs, a skill map, and verified reading.
Cloud
AZ-104 Microsoft Azure Administrator
Pass the AZ-104 cert and administer Azure — exam-domain map, flashcard decks, practice exams, free hands-on IaC/CLI/Azurite labs, a skill map, and free-first resources.
AI-300 — Microsoft ML Operations Engineer Associate
MLOps + GenAIOps on Azure (Azure ML + Foundry, IaC, CI/CD) — five skill-area decks (99 cards), an 87-question scenario practice exam, a study routine, and free-first resources, authored against the official skills-measured guide.
AWS Machine Learning Engineer – Associate (MLA-C01)
Operate ML on AWS (SageMaker-centric) — four domain-weighted decks (98 cards), an 89-question scenario practice exam, a study routine, and free-first resources verified against the official AWS exam guide.
Google Cloud Professional Machine Learning Engineer
The full ML lifecycle on Vertex AI (now including generative AI) — six section decks (99 cards), an 88-question scenario practice exam, a study routine, and free-first resources, authored against the official exam guide.
Engineering & Hardware
Electronics & PCB Design
A course, not a reading list. Ordered units from charge and Ohm's law through to board bring-up — each explains why the topic matters, sends you to specific chapters and videos with an instruction for each, and flags what those sources gloss over. Track A builds the electronics; Track B designs the board. Free sources only.
Additive Manufacturing & 3D Printing
3D printing as industry defines it — all seven ISO/ASTM 52900 process families rather than the desktop corner of one, plus materials, orientation and anisotropy, design for additive, post-processing, quality and the family-specific hazards. Aimed at the SME CAM-F body of knowledge.
Soldering & Surface Mount
The hand skill, and the knowledge that makes practising it productive. What a joint physically is, why flux is the active ingredient, packages, rework, reflow, and the IPC-A-610 acceptance classes — where 'good' is defined by class, not taste. Heavy on Skill and Risk elements that no card can close.
KiCad — PCB Layout
The tool, taught as a tool — KiCad's data model, library portability, capture and ERC, board setup before placement, the interactive router, zones, DRC and fabrication outputs. PCB design theory lives in the Electronics & PCB Design path and is assumed here.
Code-CAD with build123d
Solid modelling in Python. The B-rep kernel first, then sketches, the two APIs, locations and selectors — the two places code-CAD models silently break — and finally testing geometry in CI, which is the one thing this approach can do that GUI CAD cannot. Ends in a portfolio artifact, not a certificate.
FreeCAD & Parametric CAD
Parametric modelling discipline taught through FreeCAD — constrained sketching (the skill it lives or dies on), feature history, datum geometry, the topological naming problem in its own unit, spreadsheet-driven parameters, assemblies and a manufacturable drawing set.
Drone Technology
How unmanned aircraft work as engineered systems — not how to fly one. Airframes, propulsion, battery hazards, sensors and the state estimation everything above it depends on, flight control and tuning, radio links, autonomy, failsafes and reading a crash log.
FAA Part 107 — Remote Pilot
The commercial small-unmanned-aircraft certificate against the FAA's own knowledge areas — rules and the numbers, waivers and airspace authorisation, remote ID, sectional charts as a trained skill, airspace, weather, performance, decision-making, and night or over-people operations.
ROS 2 & Robotics
The robotics middleware in dependency order. Workspaces second because that is where beginners stall, QoS taught properly because incompatible policies fail silently, and coordinate frames at the centre because description, simulation, perception, navigation and manipulation are all wrong together if the transform tree is wrong.
Satellite Operations
Flying a spacecraft that already exists — orbital elements, regimes, perturbations, element-set propagation and its shelf life, pass prediction, spacecraft subsystems, link budgets, the ground segment, telemetry and command, anomalies and conjunctions. Ends by decoding a real pass.
Orbital Compute & Space Infrastructure
Computation in orbit as an engineering problem rather than a pitch. Ordered by constraint — environment, radiation effects on silicon, thermal rejection with no convection, power, the downlink economics the whole thesis turns on, autonomous software, and qualification. Opens and closes by examining its own premise.
Amateur Radio — General then Extra
Two tracks for the two exam elements. General is really propagation, station practice, antennas and RF safety; Extra is where the mathematics arrives — receiver dynamic range and noise figure, complex impedance, filters and oscillators, SDR, and the Smith chart. Assumes Technician is already held.