Multi-cloud orchestration for ML workloads with automatic cost optimization. Use when you need to run training or batch jobs across multiple clouds, leverage spot instances with auto-recovery, or optimize GPU costs across providers.

Battle-tested PyTorch training recipes for all domains — LLMs, vision, diffusion, medical imaging, protein/drug discovery, spatial omics, genomics. Covers training loops, optimizer selection (AdamW, Muon), LR scheduling, mixed precision, debugging, and systematic experimentation. Use when training or fine-tuning neural networks, debugging loss spikes or OOM, choosing architectures, or optimizing GPU throughput.

Fast structured generation and serving for LLMs with RadixAttention prefix caching. Use for JSON/regex outputs, constrained decoding, agentic workflows with tool calls, or when you need 5× faster inference than vLLM with prefix sharing. Powers 300,000+ GPUs at xAI, AMD, NVIDIA, and LinkedIn.

Serves LLMs with high throughput using vLLM's PagedAttention and continuous batching. Use when deploying production LLM APIs, optimizing inference latency/throughput, or serving models with limited GPU memory. Supports OpenAI-compatible endpoints, quantization (GPTQ/AWQ/FP8), and tensor parallelism.

Provides guidance for automatically evolving and optimizing AI agents across any domain using LLM-driven evolution algorithms. Use when building self-improving agents, optimizing agent prompts and skills against benchmarks, or implementing automated agent evaluation loops.

Multi-agent orchestration framework for autonomous AI collaboration. Use when building teams of specialized agents working together on complex tasks, when you need role-based agent collaboration with memory, or for production workflows requiring sequential/hierarchical execution. Built without LangChain dependencies for lean, fast execution.

High-performance vector similarity search engine for RAG and semantic search. Use when building production RAG systems requiring fast nearest neighbor search, hybrid search with filtering, or scalable vector storage with Rust-powered performance.

Guarantee valid JSON/XML/code structure during generation, use Pydantic models for type-safe outputs, support local models (Transformers, vLLM), and maximize inference speed with Outlines - dottxt.ai's structured generation library

Merge multiple fine-tuned models using mergekit to combine capabilities without retraining. Use when creating specialized models by blending domain-specific expertise (math + coding + chat), improving performance beyond single models, or experimenting rapidly with model variants. Covers SLERP, TIES-Merging, DARE, Task Arithmetic, linear merging, and production deployment strategies.

Write publication-ready ML/AI papers for NeurIPS, ICML, ICLR, ACL, AAAI, COLM. Use when drafting papers from research repos, structuring arguments, verifying citations, or preparing camera-ready submissions. For systems venues (OSDI, NSDI, ASPLOS, SOSP), use systems-paper-writing instead.

Generates optimized prompts for any AI tool. Use when writing, fixing, improving, or adapting a prompt for LLM, Cursor, Midjourney, image AI, video AI, coding agents, or any other AI tool.

Create git commits with user approval and no Claude attribution

Debug issues by investigating logs, database state, and git history

Meta-skill workflow orchestrator for bug investigation and resolution. Routes to debug, implement, test, and commit based on scope.

Search GitHub code, repositories, issues, and PRs via MCP

Interactive workspace discovery - learn what tools, workflows, agents, and hooks are available

Problem-solving strategies for source coding in information theory

System health check (MOT) for skills, agents, hooks, and memory

Formal theorem proving with research, testing, and verification phases

Code quality checks, formatting, and metrics via qlty CLI

QLTY During Development

Search past reasoning for relevant decisions and approaches

Check reference SDK implementations using btca ask

Release preparation workflow - security audit → E2E tests → review → changelog → docs

Store a learning, pattern, or decision in the memory system for future recall

Comprehensive code review workflow - parallel specialized reviews → synthesis

Security audit workflow - vulnerability scan → verification

Computational geometry with Shapely - create geometries, boolean operations, measurements, predicates

Create and configure Claude Code sub-agents with custom prompts, tools, and models

TDD workflow for migrations - orchestrate agents, zero main context growth

Test-driven development workflow with philosophy guide - plan → write tests → implement → validate

Comprehensive testing workflow - unit tests ∥ integration tests → E2E tests

Full 5-layer analysis of a specific function. Use when debugging or deeply understanding code.

Friendly onboarding when users ask about capabilities

Goal-based workflow orchestration - routes tasks to specialist agents based on user goals

Evaluate acquisition channels using unit economics, customer quality, and scalability. Use when deciding whether to scale, test, or kill a growth channel.

Diagnose SaaS business health across growth, retention, efficiency, and capital. Use when preparing a business review or prioritizing urgent fixes.

Create a company research brief with executive quotes, product strategy, and org context. Use when preparing for interviews, competitive analysis, partnerships, or market-entry work.

Run a customer journey mapping workshop with adaptive questions and outputs. Use when you need to map stages, actions, emotions, pain points, and opportunities for a persona and scenario.

Guide the PM-to-Director transition across preparing, interviewing, landing, and recalibrating. Use when leadership scope is changing and you need practical coaching.

Frame an epic as a testable hypothesis with target user, expected outcome, and validation method. Use when defining a major initiative before roadmap, discovery, or delivery planning.

Look up SaaS finance metrics, formulas, and benchmarks fast. Use when you need a quick metric definition, formula, or benchmark during analysis.

Uncover customer jobs, pains, and gains in a structured JTBD format. Use when clarifying unmet needs, repositioning a product, or improving discovery and messaging.

Guide teams through Lean UX Canvas v2. Use when framing a business problem, surfacing assumptions, and defining what to learn next.

Build an Opportunity Solution Tree from outcomes to opportunities, solutions, and tests. Use when a stakeholder request needs problem framing before you decide what to build.

Select the right Proof of Life (PoL) probe based on hypothesis, risk, and resources. Use this to match the validation method to the real learning goal, not tooling comfort.

Write an Amazon-style press release that defines customer value before building. Use when aligning stakeholders on a new product, feature, or strategic bet.

Write a user-centered problem statement with who is blocked, what they are trying to do, why it matters, and how it feels. Use when framing discovery, prioritization, or a PRD.

Structure a spoken PM product-sense answer with assumptions, segmentation, pain-point prioritization, and MVP tradeoffs. Use when practicing design, improve, or build-next interview questions.

Create a proto-persona from current research, market signals, and team knowledge. Use when you need a working customer profile before deeper validation.