Claude Skills by thedixitjain

Convert evaluation traces and production logs into SFT examples and preference pairs. Use when graded traces or failure examples exist and need to become training data, when applying rejection sampling to model outputs, or when building DPO pairs from passing and failing runs.

Regulator-grade feature attribution for any LSTM/Transformer signal — single-entry PageRank ranks the top-K features that drove the prediction (ADR-126 Phase 6, ADR-123 single-entry PR)

Detect current market regime using npx neural-trader — bull/bear/ranging/volatile classification with recommended strategy. Use when the user asks about market conditions, wants to pick a strategy for current conditions, or before running a backtest/signal that should be regime-aware.

Train neural models (LSTM, Transformer, N-BEATS) on market data using npx neural-trader with confidence intervals

'Build train machine learning models with automated workflows. Analyzes datasets, selects model types (classification, regression), configures parameters, trains with cross-validation, and saves model artifacts. Use when asked to \"train model\" or \"evalua... Trigger with relevant phrases based on skill purpose. '

Use Transformers.js to run state-of-the-art machine learning models directly in JavaScript/TypeScript. Supports NLP (text classification, translation, summarization), computer vision (image classification, object detection), audio (speech recognition, audio classification), and...

Hugging Face Transformers for loading Hub models, running pipeline inference, text generation, and Trainer fine-tuning on NLP, vision, audio, and multimodal tasks. Use when working with AutoModel, pipelines, tokenizers, or TrainingArguments—not for general ML outside the Transformers library.

'Configure triton inference config operations. Auto-activating skill for ML Deployment. Triggers on: triton inference config, triton inference config Part of the ML Deployment skill category. Use when configuring systems or services. Trigger with phrases like \"triton inference config\", \"triton config\", \"triton\". '

Train and fine-tune transformer language models using TRL (Transformers Reinforcement Learning). Supports SFT, DPO, GRPO, KTO, RLOO and Reward Model training via CLI commands.

Assess whether experiment results are credible enough to influence product decisions. Use when checking false positive or false negative risk, underpowered metrics, suspiciously large lifts, replication needs, meta-analysis, stratified sampling, covariate adjustment, or whether A/B test insights should be trusted.

'Optimize machine learning model hyperparameters using grid search, random search, or Bayesian optimization. Finds best parameter configurations to maximize performance. Use when asked to \"tune hyperparameters\" or \"optimize model\". Trigger with relevant phrases based on skill purpose. '

Use when packaging code, data, samplers, solvers, and logs for a UAI submission's 50 MB supplementary ZIP or a public post-acceptance release, making probabilistic-inference claims independently runnable while keeping every file double-blind, given that UAI reviewers may open the archive but are not obliged to read it.

Use when packaging the artifacts behind a UIST paper — code, toolkits, hardware design files, and datasets — first as anonymous review-time evidence that the system is real, then as a public release engineered for reuse, in a venue with no formal badge committee doing the checking for you.

Use when designing or auditing the evaluation of a UIST paper — choosing among technical benchmarks, controlled comparisons, usability walkthroughs, expert sessions, and demonstration applications, matching evaluation shape to the systems claim, and avoiding the ritual study that proves nothing the paper asserts.

Track physical units and propagate measurement uncertainty in scientific calculations using pint and uncertainties. Use for unit conversion and dimensional checking, GUM uncertainty budgets, Type A and Type B evaluation, coverage factors and expanded uncertainty, Monte Carlo propagation, significant-figure and plus-minus reporting, error propagation through curve fits, CODATA constants, auditing Python code for stripped units or broken uncertainty propagation, and order-of-magnitude plausibil...

Unsloth: 2-5x faster LoRA/QLoRA fine-tuning, less VRAM.

Use when packaging artifacts for USENIX Security Symposium evaluation — the mandatory Phase-1 availability check that acceptance is conditional on, the optional Phase-2 push for Artifacts Functional and Results Reproduced badges, and building security artifacts (exploits, scanners, datasets) that evaluators can run safely.

Use this skill for processing and analyzing large tabular datasets (billions of rows) that exceed available RAM. Vaex excels at out-of-core DataFrame operations, lazy evaluation, fast aggregations, efficient visualization of big data, and machine learning on large datasets. Apply when users need to work with large CSV/HDF5/Arrow/Parquet files, perform fast statistics on massive datasets, create visualizations of big data, or build ML pipelines that do not fit in memory.

Use when packaging an IEEE VIS artifact for the Graphics Replicability Stamp Initiative (GRSI) / TVCG Replicability Stamp and the IEEE VIS Open Practices program, covering what an independent GRSI volunteer reproduces first, DOI-issuing archives, evaluator-proof documentation for visualization code and data, and how VIS reproducibility differs from ACM-style artifact badges.

Use when designing or auditing IEEE VIS evaluations, covering how to match evidence to the contribution type (perceptual study, controlled user study, algorithm benchmark, design-study validation, qualitative work), controlled experiment design with power and effect sizes, CVD-safe and perceptually grounded encoding choices, task taxonomies, and provenance so a TVCG reviewer trusts the result.

Fine-tune vision-language models (VLMs) with supervised learning on image+text data. Use when adapting a VLM to a visual domain or task, configuring frozen-vision-tower LoRA, or debugging a VLM fine-tune that trains without learning.

Use when preparing a PVLDB artifact for the pVLDB Reproducibility Evaluation or the ACM availability badge, covering the mandatory participation rule for EA&B papers, the four artifact surfaces evaluators rebuild, packaging for a rerun by strangers, and positioning for the Best Reproducible Paper Award at VLDB.

Use when designing or auditing the evaluation of a VLDB paper, covering workload and dataset realism at scale, competitor tuning fairness, scalability curves versus single points, tail-latency and throughput reporting, ablations that isolate the mechanism, and the loss-case disclosure PVLDB reviewers look for first.

Use when packaging code, data, and models for a WACV paper, covering the anonymous review artifact versus the public post-acceptance release, reproducing constraint-based applications claims (latency, power, robustness) not just accuracy, dataset licensing and release, and keeping the artifact in sync across the two-round Revise-and-Resubmit lap.

Use when preparing a WACV camera-ready after acceptance, covering de-anonymization, IEEE Xplore plus CVF open-access dual publication, IEEE copyright and PDF checks, the dataset and code release obligation, per-paper registration, and preparing the winter-conference talk or poster once a paper clears Round 1 or Round 2.

'Execute wandb experiment logger operations. Auto-activating skill for ML Training. Triggers on: wandb experiment logger, wandb experiment logger Part of the ML Training skill category. Use when working with wandb experiment logger functionality. Trigger with phrases like \"wandb experiment logger\", \"wandb logger\", \"wandb\". '

Use when packaging datasets, models, or code for the Web Conference (WWW) Artifacts Available badge or for reviewer scrutiny, covering archival-repository choice, the light-verification bar, web-data licensing and takedown realities, anonymized artifacts during review, and what the badge does and does not certify.

W&B: log ML experiments, sweeps, model registry, dashboards.

Use when targeting IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) or deciding whether a computer-science manuscript fits this venue. Encodes conference fit, framing, evidence bar, submission-cycle checks, rebuttal posture, and desk-reject risks for computer vision applications.

Use when packaging code, data, and models as evidence for a WSDM paper - anonymous repositories cited in the PDF, the proprietary-log dilemma of web-scale research, public-benchmark substitution tiers, WSDM Cup datasets, and what credible artifact release looks like at a venue without a formal badge process.

Use when designing or auditing the evaluation of a WSDM paper - offline ranking and recommendation metrics with bias controls, temporal-split protocols for interaction logs, baseline selection from recent WSDM/SIGIR/KDD editions, ablations that isolate the mechanism, efficiency reporting, and online-evidence framing.

Sub-skill técnica de Yann LeCun. Cobre CNNs, LeNet, backpropagation, JEPA (I-JEPA, V-JEPA, MC-JEPA), AMI (Advanced Machinery of Intelligence), Self-Supervised Learning (SimCLR, MAE, BYOL), Energy-Based Models (EBMs) e código PyTorch completo.

'Configure CI/CD pipeline for Abridge clinical AI integrations with GitHub Actions. Use when setting up automated testing, FHIR validation, HIPAA compliance checks, or deployment pipelines for healthcare AI applications. Trigger: \"abridge CI\", \"abridge GitHub Actions\", \"abridge pipeline\", \"abridge automated testing\", \"abridge CI/CD\". '

'Implement Abridge ambient clinical documentation capture-to-note pipeline. Use when building the primary encounter workflow: audio capture, real-time transcription, AI note generation, and EHR note insertion. Trigger: \"abridge clinical workflow\", \"abridge encounter pipeline\", \"ambient documentation workflow\", \"abridge note generation\". '

'Deploy Abridge clinical AI integration to HIPAA-compliant cloud infrastructure. Use when deploying to GCP Cloud Run, AWS ECS, or Azure Container Apps with healthcare-grade secrets management and compliance controls. Trigger: \"deploy abridge\", \"abridge production deploy\", \"abridge Cloud Run\", \"abridge AWS deploy\", \"abridge HIPAA infrastructure\". '

Use when organizing appendices and supplementary material for an ACL paper under ACL Rolling Review, covering the mandatory Limitations and optional ethics sections, appendices after references, anonymized software and data archives, the no-cloud-links rule, and deciding what must stay in the 8-page or 4-page body.

Use when reasoning about the ACM MM (ACM Multimedia) review pipeline — thematic-area routing to reviewers and area chairs, the OpenReview double-blind process and its single-blind track exceptions, the optional anonymous rebuttal, the meta-review and decision, and the oral/poster and award tiers, and where an author actually has leverage.

'Deploy and orchestrate Vertex AI ADK agents using A2A protocol. Manages AgentCard discovery, task submission, Code Execution Sandbox, and Memory Bank. Use when asked to \"deploy ADK agent\" or \"orchestrate agents\". Trigger with phrases like ''deploy'', ''infrastructure'', or ''CI/CD''. '

'Execute use when provisioning Vertex AI ADK infrastructure with Terraform. Trigger with phrases like \"deploy ADK terraform\", \"agent engine infrastructure\", \"provision ADK agent\", \"vertex AI agent terraform\", or \"code execution sandbox terraform\". Provisions Agent Engine runtime, 14-day code execution sandbox, Memory Bank, VPC Service Controls, IAM roles, and secure multi-agent infrastructure. '

'Implement load testing, auto-scaling, and capacity planning for Adobe API integrations with k6 scripts targeting Firefly, PDF Services, and Photoshop APIs, plus Kubernetes HPA configuration. Trigger with phrases like \"adobe load test\", \"adobe scale\", \"adobe performance test\", \"adobe capacity\", \"adobe benchmark\". '

'Set up comprehensive observability for Adobe API integrations with Prometheus metrics, OpenTelemetry traces, structured logging, and alert rules covering Firefly, PDF Services, and Photoshop APIs. Trigger with phrases like \"adobe monitoring\", \"adobe metrics\", \"adobe observability\", \"monitor adobe\", \"adobe alerts\", \"adobe tracing\". '

'Implement Adobe I/O Events webhook registration, RSA-SHA256 signature verification, challenge handshake, and event-driven architectures with Creative Cloud, Experience Platform, and Firefly Services events. Trigger with phrases like \"adobe webhook\", \"adobe events\", \"adobe I/O events\", \"adobe event registration\", \"adobe notifications\". '

Autonomous DevSecOps & FinOps Guardrails. Orchestrates Gemini 3 Flash to audit Linux Kernel patches, Terraform cost drifts, and K8s compliance.

Reference guide for Agentica multi-agent infrastructure APIs

Set up AI Runway on AKS — from bare cluster to running model. Covers cluster verification, controller install, GPU assessment, provider setup, and first deployment. WHEN: \\\"setup AI Runway\\\", \\\"onboard AKS cluster\\\", \\\"install AI Runway\\\", \\\"airunway setup\\\", \\\"deploy model to AKS\\\", \\\"GPU inference on AKS\\\", \\\"KAITO setup on AKS\\\", \\\"run LLM on AKS\\\", \\\"vLLM on AKS\\\", \\\"set up model serving on AKS\\\", \\\"AI Runway controller\\\".

'Configure CI/CD pipeline for Alchemy-powered Web3 applications. Use when setting up automated testing with Hardhat forks, smart contract verification, or testnet deployment pipelines. Trigger: \"alchemy CI\", \"alchemy GitHub Actions\", \"web3 CI/CD pipeline\". '

'Implement Alchemy Notify webhooks for real-time blockchain event notifications. Use when tracking wallet activity, monitoring mined transactions, watching smart contract events, or building real-time dApp features. Trigger: \"alchemy webhook\", \"alchemy notify\", \"alchemy events\", \"alchemy address activity\", \"alchemy real-time notifications\". '

'Manage alertmanager rules config operations. Auto-activating skill for DevOps Advanced. Triggers on: alertmanager rules config, alertmanager rules config Part of the DevOps Advanced skill category. Use when configuring systems or services. Trigger with phrases like \"alertmanager rules config\", \"alertmanager config\", \"alertmanager\". '

'Implement Algolia indexing pipeline: data sync, partial updates, synonyms, and rules. The secondary money-path workflow: keep your index in sync with source data. Trigger: \"algolia indexing\", \"sync data to algolia\", \"algolia synonyms\", \"algolia rules\", \"algolia partial update\", \"algolia reindex\". '

'Set up observability for Algolia: Prometheus metrics for search latency/errors, OpenTelemetry tracing, structured logging, and Grafana dashboards. Trigger: \"algolia monitoring\", \"algolia metrics\", \"algolia observability\", \"monitor algolia\", \"algolia alerts\", \"algolia tracing\", \"algolia dashboard\". '