AI / ML Engineer
Build, train, and deploy machine learning models. Work with large datasets and cutting-edge AI frameworks to solve real problems.
What a typical day looks like
My day usually starts with a coffee and a quick scan of Twitter for AI papers and product announcements — the field moves so fast that missing a week feels like missing a year. Then I check our experiment tracking dashboard: did the model that trained overnight beat the current production model on our evaluation set? If yes, we promote it. If no, we figure out why. Mornings are usually for the harder thinking: designing a new feature pipeline, debugging why a model is underperforming on a subset of data, or reviewing a research paper relevant to the problem we are working on. Lunch is sometimes a team thing where we walk through an interesting paper or blog post. Afternoons are coding-heavy: training new model variants, writing data processing scripts, building evaluation harnesses. Many afternoons end with a model training run that will finish overnight, ready to evaluate tomorrow morning.
Hour-by-hour
Skills you need
Required
Nice to have
Build these to stand out
Hands-on projects beat any CV bullet point. Pick one and finish it.
RAG-Powered Q&A Bot Over Your Own Docs
Build a chatbot that answers questions about a set of documents (your CV, your notes, a textbook). Use vector embeddings, a vector DB (Chroma or Pinecone), and an LLM (OpenAI or open-source). Add a simple web UI.
Shows you understand modern AI engineering. Hottest portfolio piece in 2026.
Fine-Tuned LLM for Domain-Specific Task
Pick a domain (legal, medical, code review). Collect or generate a dataset of 500 to 5,000 examples. Fine-tune a small open-source model (Llama 3.1 8B or Phi-3) using LoRA. Evaluate against the base model. Deploy on a budget GPU (Modal, Replicate, or local).
Demonstrates real ML engineering: data, training, evaluation, deployment. Standout project.
Production ML Pipeline End-to-End
Build a complete pipeline: data ingestion (from a public API), feature engineering, model training, automated evaluation, deployment to a REST API. Add monitoring for data drift and model performance. Bonus: CI/CD with GitHub Actions to retrain weekly.
Senior-level MLE portfolio piece. Shows you can ship ML, not just train models.