AI / Machine Learning Engineer

The lingua franca of ML. Get fluent with the standard library, typing, and virtual environments before touching frameworks.

Array math and dataframes - the workhorses for loading, cleaning, and reshaping data.

Vectors, matrices, gradients, and distributions. Enough to reason about how models learn, not a pure-math degree.

Most training data starts in a warehouse. Comfortable joins and aggregations are non-negotiable.

Regression, classification, clustering, train/validation/test splits, over- and under-fitting, and evaluation metrics.

The standard toolkit for classical ML: pipelines, feature engineering, and model selection.

Framing a problem, exploratory analysis, and honest evaluation - the difference between a demo and a result.

Backpropagation, CNNs, RNNs, and the transformer architecture that underpins modern AI.

A deep-learning framework end to end - datasets, training loops, and GPU acceleration.

Load, fine-tune, and evaluate pretrained models with the Transformers ecosystem.

How large language models work, context windows, and prompt engineering for reliable outputs.

Ground models in your own data with embeddings, a vector store, and retrieval-augmented generation.

Tool use and multi-step agents with frameworks like LangChain / LangGraph.

Adapt a base model to a task with full or parameter-efficient fine-tuning (LoRA/QLoRA).

Experiment tracking, model registries, deployment, monitoring, and retraining - shipping models, not notebooks.

Package and serve models on AWS/GCP with Docker and Kubernetes.

Low-latency, cost-aware inference - APIs, batching, and (for LLMs) tools like vLLM.