Library
Page 2 of 14
🦀Rust Borrow Checker Deep Dive
Build a precise mental model of Rust's borrow checker — from the shared-XOR-mutable invariant through NLL and reborrows — so you can refactor tangled code into a clean version the compiler accepts without reaching for clone().
💡Run AI-Augmented Brainstorming Sessions
Stop letting ChatGPT collapse your brainstorms into faster bad decisions. This path teaches the Double Diamond — AI for divergence, you for convergence — so you walk out of every session with more options and a sharper pick.
🛰️Learn the Stoic View From Above
Turn the Stoic view from above into a disciplined 4-step zoom-out you run on a real frustration every day. By day 14 you'll have a recorded personal script you can replay whenever perspective collapses.
🏛️Learn the Stoic Practice of Negative Visualization
Turn a Stoic exercise into a bounded 3-minute ritual that lands as gratitude, not dread. By day 14 you'll have a monthly loss-audit you actually run on the first Sunday of every month.
🎭Learn Iambic Pentameter in Poetry
Stop nodding at 'da-DUM da-DUM' and actually hear it. Train your ear to scan Shakespeare, spot the substitutions poets use for effect, and write your own iambic pentameter couplets and quatrains with confidence.
🏛️Learn Amor Fati: Love of Fate
Turn 'love of fate' from a poetic line into a daily 3-sentence reframe and a Sunday review you actually run — so when bad things hit your week, you have a concrete move instead of a vague aspiration.
📈Know When Not to Use ML for Time-Series Forecasting
Stop reaching for LSTMs on tiny series — enforce the baseline ladder (naive, seasonal-naive, ARIMA, then ML), backtest each one properly on real data, and write the decision criteria your team will use to escalate to ML only when the simple model is genuinely beaten.
🇳🇱Dutch Word Order and V2
Stop translating German V2 into Dutch and start hearing Dutch word order on its own terms. Learn where the conjugated verb has to live, what 'fronting' really moves, how omdat-clauses differ from weil-clauses, and finish by writing short paragraphs with fronted elements and subordinate clauses your Dutch friends won't have to mentally re-parse.
🇳🇱Dutch Modal Particles (wel, maar, toch, even)
Stop sounding like a textbook in Dutch. Master wel, maar, toch, and even to add the warmth, softness, and shared feeling that native speakers hear in every sentence.
📋Document Datasets with Datasheets
Datasets get retrained; the quirks get rediscovered. Walk the Gebru et al. datasheet section by section against a real dataset, compare it to model cards and Google's data cards, then audit one of your team's datasets and flag the gaps.
📉Detect Anomalies in Time-Series Data
Stop alerting on every weekend dip and missing the real incidents — learn to separate point, contextual, and collective anomalies, match each to the right detector (Z-score, isolation forest, LSTM autoencoder), and design an SLI alerting rule that survives weekly seasonality plus a slow trend.
🧠Build an AI-Augmented Personal Knowledge Base
Pick the three jobs AI is actually good at inside a note vault — connection discovery, cross-note Q&A, and atomic distillation — then design a daily workflow that still works when the model is offline.
⚖️Build a Mental Model of the EU AI Act's Risk Tiers
Build the four-tier mental model of the EU AI Act — unacceptable, high, limited, minimal — through worked examples, then self-classify your own product and write a one-page tier assessment you could defend to outside counsel.
📇Write Model Cards for AI Transparency
Stop writing model cards from memory. Walk every section of the Mitchell et al. card with a worked classifier, then critique a real card from a major lab and name what's missing — so transparency becomes a habit, not a deliverable.
🔀Use Query Expansion to Improve RAG Recall
Compare four query-expansion patterns — synonym, multi-query, step-back, and HyDE — on the same hard query so each one's strength is visible, then design a query-expansion stage for a customer-support RAG with 30% short queries.
🏷️Use Metadata Filtering in Vector Search
Contrast pre-filter, post-filter, and partitioned-index strategies for metadata-aware vector search on the same dataset so the recall failure mode becomes visible, then design a metadata schema for a multi-tenant RAG that needs sub-100ms queries.
🧪Use Eval Frameworks: Ragas, DeepEval, TruLens
Stop hunting for a single 'best' RAG eval tool. You'll learn the four core RAG metrics, score the same app in Ragas and DeepEval, see where each framework wins, and ship a layered eval stack you can defend to your team.
🧹Use AI to Refactor Legacy Code
Stop shipping AI-refactored legacy code that subtly breaks behavior. By the end you'll take a 200-line legacy function through explore → characterize → refactor → review and produce a version with provable behavior preservation — using AI on the careful steps, not as a shortcut around them.
🚦Use AI Gateways: OpenRouter, Portkey, Helicone
Stop choosing a gateway because a blog post said so. By the end you can pick OpenRouter, Portkey, Helicone, or self-host for a real multi-region app and defend it on failover, cost, and observability.
🎙️Understand Speech-to-Text Accuracy and WER
Stop trusting WER numbers from someone else's benchmark — build a 50-clip eval set from your own production audio so the next time you swap transcription vendors, the decision rests on your data, not theirs.
🌐Understand Multilingual Embeddings
Stop bolting translation onto English-only RAG. By the end you'll understand how knowledge distillation aligns embedding spaces across languages — and you'll have a concrete plan for support-doc search across 12 languages, with the low-resource gotchas mapped before you ship.
🖼️Understand Image Embeddings and Visual Search
Bridge from text embeddings to image embeddings, then design a duplicate-photo finder for your own library — without ever reaching for perceptual hashes.
🖼️Understand CLIP and Contrastive Image-Text Learning
Stop treating CLIP as a black-box embedding API. By hand-building the contrastive matrix on five image-caption pairs and tracing one shared embedding space, you'll design a 'photo of a red bicycle' search over an unlabeled folder — and know exactly why it works.
🔏Understand C2PA Content Credentials and AI Watermarking
Stop treating 'is this AI?' as a vibes question. Separate the three layers — hash (hard binding), watermark (soft binding), signed manifest — so provenance becomes verifiable evidence, then design a flow for a media product that ships both human and AI content.