Results for “dual-track”

Use when planning an IPSN-lineage project timeline from track/venue selection through the CPS-IoT Week deadline, double-blind submission, deployment and hardware logistics, rebuttal, the Best Research Artifact Award, and the dual ACM/IEEE camera-ready — with honest handling of the fact that IPSN merged into SenSys.

Two-party working posture — user as director, agent as executor. Every fork, tradeoff, and taste choice is surfaced via batched AskUserQuestion with structural framing, a recommended default, and concrete previews when comparison is visual, so the human steers direction while the agent handles implementation. Eliminates the review-bottleneck (no giant diff to approve at the end — review is distributed across picks) and prevents codebase-understanding debt (the user remembers the architecture because they picked it). Use whenever the user invokes /duet, or says "work with me", "ask before", "check with me", "I want to decide", "don't assume", "human-in-the-loop", "co-author", "pair with me", "duet", or whenever a task clearly involves aesthetic, architectural, or irreversible strategic decisions — even without those exact words. Pair with the Duet output style to minimize cognitive load between picks.

Track and visualize ML training experiments with Trackio, including logging metrics, firing alerts, and retrieving data via CLI. Supports real-time dashboards, webhook alerts, and HF Space syncing.

Select alternative execution paths when a plan step fails — branch on outcomes instead of blind retry. Load when a structured plan step fails, an unexpected observation arrives, or the user asks what to try next after an error. Also triggers on "try another approach", "route around this failure", "replan on failure", "if X fails try Y", or outcome-based branching during multi-step work. Pairs with structured-planning and debug-and-fix. Does not replace root-cause debugging — adds plan-level path selection.

Establishes a two-party working posture where the user directs and the agent executes, surfacing every genuine fork as a batched single-select question with a recommended default.

Claude Code + Codex dual orchestration for hybrid interactive and parallel workflows

Multi-tiered memory management (HOT/WARM/COLD) for context compaction. Invoke ONLY for explicit compaction events: post-`/compact` cleanup, MEMORY.md tier promotion, archive batch, or "trim my context". NOT for general recall (use deep-recall) or routine memory writes (use storage-router). Triggers: "compact memory", "promote to durable", "archive old context", "tier this".

Composite skill — safely refactor a module end-to-end with sequencing, parallel implementation, post-refactor cleanup, and rationale capture. Chains refactor-plan (phased plan + rollback) → three-man-team (architect/builder/reviewer in parallel) → fix-the-suite post-refactor → adr-write → docs-sync. Use for non-trivial refactors that need both careful sequencing and durable record.

Trains large language models (2B-462B parameters) using NVIDIA Megatron-Core with advanced parallelism strategies for maximum GPU efficiency.

Track ML experiments with automatic logging, visualize training in real-time, optimize hyperparameters with sweeps, and manage model registry with W&B.

Workflow for private repos + Colab training. Trigger when: (1) training on Colab with private repo, (2) separating dev from production, (3) avoiding untested code in production.

Scan codebases for technical debt, score severity, track trends, and generate prioritized remediation plans. Use when users mention tech debt, code quality, refactoring priority, debt scoring, cleanup sprints, or code health assessment. Also use for legacy code modernization planning and maintenance cost estimation.

Parallel execution engine for high-throughput task completion

Scan codebases for technical debt, score severity, track trends, and generate prioritized remediation plans.

Parallel execution engine for high-throughput task completion

Review AI agent skills using a dual-axis method: deterministic code-based checks and LLM deep review, with weighted scoring and improvement recommendations.

Trains large language models (2B-462B parameters) using NVIDIA Megatron-Core with advanced parallelism strategies. Use when training models >1B parameters, need maximum GPU efficiency (47% MFU on H100), or require tensor/pipeline/sequence/context/expert parallelism. Production-ready framework used for Nemotron, LLaMA, DeepSeek.

Track ML experiments with automatic logging, visualize training in real-time, optimize hyperparameters with sweeps, and manage model registry with W&B - collaborative MLOps platform

Audits UX quality of a codebase or validates implementation against design mockups, then fixes issues and commits changes.

Train Mixture of Experts (MoE) models using DeepSpeed or HuggingFace. Use when training large-scale models with limited compute (5× cost reduction vs dense models), implementing sparse architectures like Mixtral 8x7B or DeepSeek-V3, or scaling model capacity without proportional compute increase. Covers MoE architectures, routing mechanisms, load balancing, expert parallelism, and inference optimization.

Coordinate multiple subagents/worktrees for parallel workstreams. Decomposes larger tasks into independent sub-tasks, dispatches each to a dedicated agent.

Parallel execution engine for high-throughput task completion

Analyzes and documents a two-track design system with cinematic dark marketing pages and light transactional surfaces, including color palettes, typography hierarchy, and component vocabulary.

Produce displace/team/ghost talk tracks for the incumbent on a recompete pursuit. Use when user wants competitive angles saved to the pursuit vault; optional multi-turn LLM for customer-facing phrasing.

Use this skill when creating, managing, or working with Conductor tracks - the logical work units for features, bugs, and refactors. Applies to spec.md, plan.md, and track lifecycle operations.

Wire ETL, ingestion, cron, edge-function, and queue jobs correctly. Use for "build a pipeline", "sync X into Y", "nightly aggregation", "cron double-counts", "dedupe", "backfill", "the numbers are wrong after a retry". Bakes in idempotency, atomic writes, data contracts, dead-letter, and observability.

v5.6.0 joint multi-TF model: single model per symbol with broadcast 1Hour context replaces dual 15Min/1Hour models. Trigger: (1) replacing weighted-voting model aggregation, (2) adding broadcast features to vectorized env, (3) limited training data + worried about overfitting from doubling obs_dim, (4) backtest builder mismatch with newer feature counts.

Specialist orchestrating codex exec / Antigravity CLI through dual paradigms — COMPETE (multi-variant comparison, select best) and COLLABORATE (decompose tasks across engines, integrate). Supports Solo/Team/Quick execution modes.

Track multi-step work with todo, plan mode, kanban, and missions, each suited to different lifespans.

Scan codebases for technical debt, score severity, track trends, and generate prioritized remediation plans. Use when users mention tech debt, code quality, refactoring priority, debt scoring, cleanup sprints, or code health assessment. Also use for legacy code modernization planning and maintenance cost estimation.

Track ML experiments with automatic logging, visualize training in real-time, optimize hyperparameters with sweeps, and manage model registry with W&B - collaborative MLOps platform

Track ML experiments with automatic logging, visualize training in real-time, optimize hyperparameters with sweeps, and manage model registry with W&B - collaborative MLOps platform

Validate and configure packed sequences and long-context training in Megatron-Bridge, distinguishing offline packed SFT for LLMs from in-batch packing for VLMs with correct context parallelism constraints.