Machine Learning

Build clinical prediction models on EHR data with MIMIC, eICU and OMOP datasets

Load Therapeutics Data Commons datasets with scaffold splits and standard benchmarks

Train random forests and rank variable importance with error and importance plots

Fit Cox, random survival forest and survival SVM models with C-index and Brier scores

Explain black-box model predictions with SHAP values, dependence and fairness checks

Rank features with SVM-RFE on expression matrices and pick a count from CV error

Build graph neural networks for molecules and proteins in PyTorch

Train XGBoost models on tabular data with importance rankings and plots

Design, train and audit deep learning pipelines for imaging with proper validation

Implement attention variants from papers and wire them into a transformer stack

Define custom distance metrics for DBSCAN and scipy clustering functions

Implement and debug GRPO reinforcement learning for language model training

Use JAX for array math, autodiff, JIT compilation and scan-based RNNs

Design end-to-end ML pipelines for data prep, training, deployment and monitoring

Implement manifold-constrained hyper-connections for more stable deep network training

Ship production ML models with PyTorch, serving, feature engineering and monitoring

Build ML pipelines, experiment tracking and model registries with MLflow and Kubeflow

Train a GPT-2 scale model on one GPU with mixed precision and modern optimizers

Replay LLM request traces against a KV prefix cache to measure hit rates

Diagnose stalled reward, NaN gradients and loss issues in RL fine-tuning

Navigate the Hugging Face TRL library's trainers and shared utilities

Generate importable QuantMind model training config files in YAML or JSON

Install and troubleshoot FinBERT Chinese financial news sentiment scoring

Run an end-to-end A-share alpha factor mining pipeline with RD-Agent and backtests

Run HuggingFace models for text classification, embeddings and summarization

Fine-tune models on Microsoft Foundry with SFT, DPO or RFT and evaluate them

Set up ML experiment tracking, reproducible datasets and model readiness checks

Build machine learning and inference code in Rust with burn, candle and tch-rs

Choose and debug fine-tuning methods such as SFT, LoRA, DPO and GRPO by reward shape

Train, serve and evaluate machine learning models, including embeddings

Rename a label across all XML annotation files in a folder

Move images that lack an XML annotation file into a separate folder

Move JSON label files with empty shapes into a separate folder

Add distributed and mixed-precision training to PyTorch with Hugging Face Accelerate

Apply proven PyTorch training recipes with optimizers, schedules and mixed precision

Generate a Python experiment code skeleton from a research blueprint

Fine-tune large LLMs efficiently with LoRA, QLoRA and other adapter methods

Process large datasets for ML with streaming Ray Data pipelines across CPUs and GPUs

Speed up LLM fine-tuning with Unsloth and cut GPU memory use with LoRA

Use MLX Swift arrays, lazy evaluation and autodiff for Apple Silicon machine learning

Run LLMs and vision models on Apple Silicon with MLX Swift, including LoRA tuning

Remove refusal behavior from open-weight LLMs using interpretability methods

Guide Fully Sharded Data Parallel training in PyTorch with mixed precision and offload

Train and compare classifiers on tabular data with cross-validation in MATLAB

Engineer and select the best features for classification or regression in MATLAB

Import PyTorch, ONNX and Keras models into MATLAB as dlnetwork objects

Explain which predictors drive a tabular ML model and why it made one prediction

Clean signal data and build labeled datastores for analysis and ML in MATLAB

Train, evaluate and export neural networks in MATLAB with the trainnet API