🧭 Find your fit first
“AI job” is not one job. These roles want different skills, different backgrounds, and different amounts of maths. Find the one that fits you before you pick a course.
Builds and ships models as working software. The most common AI engineering job in India.
Day to day: Turning a model that works in a notebook into a service that works at scale, and keeping it working.
Builds products on top of existing foundation models rather than training new ones. The fastest-growing entry point.
Day to day: Designing RAG systems, agents and LLM features — prompting, retrieval, evals, guardrails and cost control.
Answers business questions with data and statistics. Closer to analysis and decisions than to shipping software.
Day to day: Framing the question, finding the data, modelling it, and — the part people underrate — explaining it to non-technical people.
Keeps models running reliably in production. A natural switch for DevOps and backend engineers.
Day to day: Pipelines, deployment, monitoring, drift detection, retraining and infrastructure cost.
Builds the pipelines every AI system depends on. Chronically in demand and often overlooked.
Day to day: Moving data reliably from where it is produced to where models can use it, at volume.
Specialises in language — search, extraction, classification, translation.
Day to day: Building systems that read and understand text, increasingly on top of LLMs.
Works on images and video — detection, recognition, inspection, medical imaging.
Day to day: Training and deploying vision models, often onto constrained or edge hardware.
Advances the state of the art. Almost always needs a postgraduate degree — be clear-eyed about that.
Day to day: Reading, proposing, experimenting and publishing. The smallest and most competitive slice of the field.
