Turn business data into answered questions with agents that define the metric, pull it consistently, check it against a second source, and state what it does not prove.
Interview as an Amazon-style bar raiser who holds an independent veto and judges each candidate against the long-term hiring bar.
Manage every resource through RAII with the rule of zero, correct smart-pointer selection, and move semantics.
Use more than one model or tool deliberately, playing to their differences rather than switching at random. Use when one model is consistently weak on part of your workflow.
Organize writing so the reader is carried from point to point without friction: logical order, transitions, and signposting.
Split data before looking at it, respect time and groups, and quarantine the test set so scores mean something.
Ship code behind flags without accumulating flag rot, through disciplined naming, expiry, and cleanup sweeps.
Build a custom-property token system with primitive and semantic tiers that supports runtime theme switching.
Decide what to hand to an agent, what to keep with a human, and what to split, based on reversibility and consequence rather than capability alone.
Work with bytes, endianness, alignment, and binary formats so data written by one system is read correctly by another.
Capturing and streaming database changes in real-time using Debezium, Kafka, and event-driven patterns for data synchronization.
Have agents argue opposing positions from the same evidence, then synthesize, so the strongest case for each option is heard before deciding.
Identify repetitive manual operational work and eliminate it, so capacity goes to improvement rather than maintenance.
Understand which costs are fixed, variable, and step, so you know how profit responds to volume and where cuts actually help.
Decide how much reliability to buy, given that each additional nine costs disproportionately more than the last.
Sequence work by dependency to find which tasks actually determine the finish date, and manage those. Use when a project has many parallel workstreams and an immovable date.
Write the repository instruction file that coding agents read, so they follow project conventions without being told each session.
Organize Go packages by domain with internal boundaries, consumer-side interfaces, and structure that grows on demand.
Run a public forum where users help each other, with seeding, moderation, and staff presence that keeps answers accurate.
Size functions to one clear job, extracting when they blur and leaving them long when the logic is linear. Use when a function grows hard to name or scan.
Instrument mobile apps for crash-free rate, ANR and hang detection, and release health you can gate rollouts on.
Choose batch size understanding its effect on gradient noise, memory, throughput, and generalisation, and adjust the learning rate with it.
Choose names that carry meaning so readers grasp intent without chasing definitions. Use when naming variables, functions, types, or files, or when a name reads as vague.
Use low- and high-fidelity wireframes and prototypes to test ideas cheaply before building. Use when exploring a design, aligning stakeholders, or validating a flow before…
Write onboarding docs that get a new contributor to a verified working setup and a first contribution fast.
Write an Amazon six-pager, a narrative memo read in silence at the start of a meeting so a decision rests on reasoning rather than a presenter's charisma.
Select Java collections by access pattern, use streams where they clarify, and default to immutability and records.
Structure async code so the event loop stays free, slow producers apply backpressure, and blocking calls never stall unrelated work.
Give feedback that improves performance, timed and framed so it is used rather than defended against. Use when coaching, reviewing work, or mentoring.
Choose a licence, comply with the licences you depend on, and keep attribution correct as a project grows.
Run transcoding, thumbnailing, and extraction as queued background work with progress, retries, and failure visibility.
Buy traffic profitably by measuring against payback and lifetime value rather than cost per click. Use when considering paid channels or when spend is rising without profit.
Reason correctly about probability, base rates, conditional probability, and expected value, avoiding the common intuition traps.
Apply normal forms pragmatically and denormalize deliberately, weighing update anomalies against read performance.
Operate as a developer relations (DevRel) engineer who closes the loop between external developers and the product team.
Persist facts about a user across sessions with provenance, correction, and expiry, so an assistant improves rather than accumulating stale assumptions.
Get talks accepted and deliver them well through audience-first proposals, narrative structure, and rehearsed, demo-safe delivery. Use when proposing or preparing a technical talk.
Replace opaque boolean flags with named enums or split functions so call sites read without the definition open.
Design autoscaling on the right metric with velocity controls, warm capacity, and flap prevention. Use when configuring autoscaling or diagnosing oscillation, lag, and cost spikes…
Configure Kubernetes workloads with correct requests, probes, disruption budgets, and workload types.
Pair a product-manager agent with engineer agents through spec, pushback, scope negotiation, and acceptance so the build matches intent.
Patterns and strategies for cache invalidation - one of the two hardest problems in computer science.
Choose between active-passive and active-active multi-region architectures with eyes open to data, cost, and failover reality.
Use .NET's built-in DI with correct lifetimes, the options pattern, and constructor injection free of service-locator drift.
Decide rollback versus fix-forward, handle migrations safely, and rehearse rollbacks so they work under pressure.
Make immutability the default so aliasing bugs turn into compile errors or no-ops instead of action at a distance.
Assess a high-risk processing activity before it ships, documenting risks, mitigations, and the decision.
Translate company objectives into desk-level goals and weekly work, keeping the line from a task to the objective visible.
Send at times that respect the recipient's local hours and working pattern, deferring anything non-urgent. Use when notifications reach people at night or on weekends.
Configure tsconfig.json deliberately: the flags that matter, module and target settings, and build vs typecheck configs.
Return errors an agent can act on, distinguishing retryable failures from permanent ones and never leaking internals. Use when building MCP tools that will fail in production.
Keep TODO comments actionable by requiring an owner, a ticket, and an expiry, or deleting them. Use when writing, reviewing, or sweeping in-code TODO, FIXME, and HACK markers.
Wire a static analysis scanner into CI so it blocks real security bugs while staying under a defined noise budget.
Provide filters that narrow results honestly, with counts that reflect what is actually available and a clear way back out.
Insert or update in one statement without races or lost updates, using the engine's conflict handling rather than check-then-write.
Plan content that reaches buyers at the moment they are looking, with a cadence you can sustain and a way to tell whether it worked.
Separate planning from execution and replan when reality diverges, so a long task does not follow a plan that stopped being true.
Choose threads, processes, or asyncio from workload shape, and use concurrent.futures without deadlocks.
Keep client state current with server changes using subscriptions, deltas, and resync, without drift or memory growth.
Pair every generating agent with an independent verifier whose only job is to refute the output, so plausible-but-wrong work is caught before it ships.