Model Training & Fine-tuning Agent Skills

Reduces peak GPU memory in Megatron Bridge training by applying expandable segments, parallelism resizing, activation recompute, and CPU offloading constraints.

Indexes Megatron Bridge recipes and recommends the best starting config based on model, GPU count, and training goal.

Run Megatron-LM (MLM) and Megatron Bridge training with mock or real data, covering correlation testing, available recipes, and multi-GPU examples.

Configure and validate CPU offloading for Megatron Bridge training, including activation offloading and optimizer state offloading with HybridDeviceOptimizer.

Trains, evaluates, exports, and runs inference for FastFoundationStereo (FFS) stereo depth estimation models, a distilled variant of FoundationStereo with lower latency.

Create and modify NeMo AutoModel training and evaluation recipes, including YAML structure, builders, and execution flow.

Identifies the weakest samples per ground-truth label in NVIDIA TAO VCN Classify experiments by running a Docker container that performs threshold sweep, weakness scoring, and per-lighting expansion, then surfaces top-K weak samples for downstream augmentation or relabeling.

Provides guidance for training Mixture-of-Experts models with long context windows, covering context parallelism sizing, selective recomputation, dispatcher choices, and practical patterns from recent experiments.

Provides practical guidance for training Mixture-of-Experts Vision-Language Models in Megatron Bridge, comparing FSDP and 3D-parallel approaches with lessons from recent multimodal experiments.

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.

Turn model work into a production ML system with data contracts, repeatable training, measurable quality gates, deployable artifacts, and operational monitoring.

Provides idiomatic PyTorch patterns and best practices for building robust, efficient, and reproducible deep learning applications, covering model architecture, training loops, data pipelines, and checkpointing.

Provides an adaptive methodology for adding machine learning models to existing codebases, covering problem framing, data readiness, architectural decoupling, and baseline model integration.

Design, simulate, and run quantum circuits on quantum computers and simulators using Google's Cirq framework.

Build, optimize, and execute quantum circuits using Qiskit on simulators or real quantum hardware from IBM, IonQ, and Amazon Braket.

Provides expert guidance on modern Julia 1.10+ development, covering performance optimization, multiple dispatch, tooling, testing, and production-ready practices.

Guides AI agents to apply advanced mathematical and probabilistic techniques (Bloom filters, HyperLogLog, FFT, etc.) for large-scale data problems where classical algorithms are optimal but math offers better asymptotic bounds.

Calibrate a new dataset from pre-recorded video files via the AutoMagicCalib REST API.

Trace and interpret the Pareto frontier across competing objectives using repeated single-objective cuOpt solves (weighted-sum and ε-constraint).

Search models, datasets, Spaces, and papers on Hugging Face Hub, retrieve repository details and documentation, run compute jobs, and use Gradio Spaces as AI tools via MCP server tools.

Estimates GPU memory required to load Safetensors or GGUF model weights for inference from the Hugging Face Hub using HTTP Range requests, without downloading weights locally.

Guides technology selection and implementation of AI and ML features in .NET 8+ applications using ML.NET, Microsoft.Extensions.AI, Microsoft Agent Framework, GitHub Copilot SDK, ONNX Runtime, and OllamaSharp.

Automate BigML machine learning operations through Composio's toolkit via Rube MCP, including model creation, training, and deployment.

Perform time series machine learning tasks including classification, regression, clustering, forecasting, anomaly detection, segmentation, and similarity search using a scikit-learn compatible Python toolkit.

Design, simulate, and run quantum circuits on Google Quantum AI hardware and partner backends using Cirq.

Build, fit, validate, and compare Bayesian models using PyMC's modern API, including hierarchical models, MCMC sampling, variational inference, posterior predictive checks, and model comparison.

Deploy and serve AI/ML models on Modal's serverless cloud platform with on-demand GPUs, autoscaling containers, persistent storage, and scheduled jobs.

Solve single and multi-objective optimization problems using NSGA-II/III, MOEA/D, and other evolutionary algorithms with customizable operators, constraint handling, and benchmark problems.

Train unsupervised machine learning models on genomic interval data from BED files, including region embeddings, single-cell ATAC-seq analysis, and consensus peak building.

Analyze whole-slide pathology images with Python: load 160+ slide formats, preprocess H&E stains, segment nuclei, construct spatial graphs, train ML models, and process multiplex immunofluorescence data (CODEX, Vectra).

Build and execute quantum circuits on IBM Quantum hardware, simulators, and third-party providers using the Qiskit framework.

Estimate cell state transitions from unspliced/spliced mRNA dynamics using scVelo, infer trajectory directions, compute latent time, and identify driver genes in single-cell RNA-seq data.

Predict molecular properties, train graph neural networks, and run drug discovery workflows using DeepChem's featurizers, models, and MoleculeNet benchmarks.

Process whole slide images for digital pathology: detect tissue, extract tiles, and prepare datasets for deep learning pipelines.

Build clinical deep-learning pipelines with PyHealth: load EHR, signal, and imaging datasets, define prediction tasks, instantiate models, train with the PyHealth Trainer, and compute clinical metrics.

Train quantum circuits like neural networks with automatic differentiation, device-independent programming, and integration with PyTorch or JAX.