dgallitelli

Use when extracting tables from PDF documents — works for digital and OCR'd PDFs, multi-column layouts, and rotated tables. Triggers on requests to "pull tables out of a PDF", "extract a table from a .pdf file", or after pdfplumber/Camelot return empty results.

I can help you with stuff

Use when reviewing, auditing, or about to publish a Claude / Agent Skill (SKILL.md plus any supporting files) to verify it follows the Agent Skills specification and Anthropic's authoring best practices. Triggers on phrases like "review this skill", "is my skill ready", "lint my SKILL.md", "audit before publishing", or before merging skill PRs.

Reviews REST API designs for consistency, naming conventions, versioning strategy, and error handling patterns

Guides agents through auditing project dependencies for known vulnerabilities, outdated packages, and license compliance issues

Guides agents through test-driven development using the red-green-refactor cycle before writing implementation code

Use when the user asks about competitive Pokemon, VGC teambuilding, metagame analysis, Pokemon sets, EV spreads, damage calculations, team synergy, matchup analysis, tournament preparation, or anything related to Pokemon doubles battles. Triggers on: "team", "VGC", "Pokemon", "metagame", "meta", "matchup", "synergy", "EV spread", "damage calc", "teambuilding", "team preview", "lead", "tournament", "regulation", "format", "counter", "check", "set", "moveset", "ability", "item", "nature", "Trick Room", "Tailwind", "weather", "terrain", "Fake Out", "Protect", "speed control", "restricted", "rental", "replica", "Pikalytics", "Showdown", "pokepaste", "ladder", "Tera", "Terastallize", "priority", "core", "archetype", "goodstuff", "sun team", "rain team", "sand", "Intimidate", "Follow Me", "Helping Hand", "redirection", "bring 4", "Bo3", "Bo1", "open teamsheet", "closed teamsheet", "Champions", "pivot", "switching", "IVs", "spread move", "restricted legendary"

Use when the user asks about Pokemon TCG rules, deckbuilding, card legality, meta decks, purchasing advice, or collection management. Triggers on: deck, TCG, Pokemon card, Standard format, Expanded, trainer card, energy, regulation mark, banned, rotation, meta deck, archetype, singles, sealed product, deckbuilding, collection, PTCGL, Limitless, prize cards, evolution, VSTAR, ex, V, GX

Standard Operating Procedure (SOP) for training and fine-tuning LLMs on Amazon SageMaker. Use when the user wants to: fine-tune, continued pretraining, CPT, preference optimization, DPO, RLHF, LoRA, QLoRA, Spectrum, full fine-tuning, SFT on SageMaker. Also use when choosing between SageMaker Training Jobs vs HyperPod clusters, selecting training containers (DLC), or GPU vs Trainium instances. Generates runnable notebooks or Python scripts. Triggers: "train llm", "fine-tune", "sagemaker training", "lora training", "qlora", "hyperpod training", "trainium training", "dlc container", "training job".

Amazon SageMaker HyperPod expert for ML training clusters with Trainium or GPU. Use when: creating HyperPod clusters, running distributed training, configuring EKS or Slurm orchestration, troubleshooting cluster issues, checking quotas, or when user mentions "hyperpod", "hyp", "ml-cluster", "trainium", "trn1", "distributed training", or "multi-node training".

Comprehensive SOP for Amazon SageMaker AI — SDK v3 code generation, serverless model customization (SFT, DPO, RLVR, RLAIF), LLM inference endpoints, model training (classical ML + LLM fine-tuning), HyperPod clusters (training + inference), Model Monitor, and AutoML with AutoGluon. Generates correct, working code every time. Triggers: "sagemaker", "training job", "deploy model", "endpoint", "llm inference", "vllm", "model customization", "djl", "fine-tune", "qlora", "lora", "rlvr", "rlaif", "dpo", "sft", "reward function", "hyperpod", "model monitor", "data quality", "bias", "autogluon", "jumpstart", "deploy on hyperpod", "training script", "inference script".

AWS Forward-Deployed Engineer (FDE) for AI/ML solutions. Builder-first technical partner who creates working prototypes, not just advice. Use when: building POCs or demos for AI/ML projects, designing architectures with Bedrock/SageMaker/AI Services, creating technical validation artifacts, or needing rapid implementation with AWS best practices. Trigger when user mentions "build me", "prototype", "POC", "demo", "FDE", or wants hands-on AI/ML building help rather than advisory. Integrates with AWS cost optimization and CDK skills for cost analysis and infrastructure.

Orchestrates tight-loop SageMaker training iteration using managed warm pools. Ensures pool reuse by enforcing matching constraints, verifying warm pool status between submissions, using persistent cache, and releasing instances on completion. Triggers: "iterate on training", "debug training job", "warm pool", "resubmit training", "bump dependency and retrain", "recipe iteration", "keep_alive", "dependency cascade", "sweep", "try different learning rates", "grid search", "train until metric", "iterate until accuracy", "train until loss", "try these hyperparameters", "bump and retry", "keep training until", "run training with different". Also auto-activates when a request implies 2+ sequential SageMaker training jobs on the same instance type — even without explicit mention of warm pools.

Discovers governed data assets in AWS catalogs (DataZone/Glue), generates SQL from natural language grounded in actual schemas, and executes governed queries via Athena with Lake Formation permissions. Use when the user says "find data", "query the catalog", "what tables do we have", "analyze this dataset", or asks about data in their lakehouse.

Create, migrate, run, debug and optimize genomics workflows in AWS HealthOmics. Use when working with bioinformatics pipelines supporting WDL, Nextflow, or CWL languages. Triggers: creating new genomics/bioinformatics workflows, migrating Cromwell/on-premises WDL pipelines, running and monitoring workflow executions, debugging failed runs, optimizing performance, setting up ECR pull-through caches, integrating from Git repos, or searching for genomics files. Keywords: healthomics, WDL, CWL, Nextflow, workflow, genomics, bioinformatics, pipeline, RNA-seq, DNA-seq, variant calling, scRNA-seq, omics.

Use when writing Python code that imports sagemaker, uses SageMaker training/serving/pipelines, or migrating from SDK v2. Triggers on import sagemaker, ModelTrainer, ModelBuilder, Pipeline, sagemaker.train, sagemaker.serve, sagemaker.mlops, SFTTrainer, DPOTrainer, Torchrun, InferenceSpec, xgboost sagemaker, sklearn sagemaker, pytorch sagemaker, deploy LLM, vLLM, DJL, HyperPod inference, model monitor, fine-tune LLM, QLoRA, training recipes, sagemaker endpoint, inference script. Also use when web search returns v2-era import paths like sagemaker.modules, sagemaker.estimator, or sagemaker.workflow.

Discovers governed data assets in AWS catalogs (DataZone/Glue), generates SQL from natural language grounded in actual schemas, and executes governed queries via Athena with Lake Formation permissions. Use when the user says "find data", "query the catalog", "what tables do we have", "analyze this dataset", or asks about data in their lakehouse.