Advances the state of the art. Almost always needs a postgraduate degree — be clear-eyed about that.
A research scientist advances the state of the art in AI by proposing, experimenting with and publishing novel techniques. This is the most academically demanding role in AI — almost all positions require a Master's degree, and most require a PhD. Be clear-eyed about this requirement before committing to this path.
Research scientists at companies like Google DeepMind, Meta FAIR, Microsoft Research and Anthropic work on fundamental problems: improving model architectures, training efficiency, alignment, reasoning, and safety. In India, research roles exist at large tech companies, research labs and increasingly at well-funded AI startups.
Support research projects, implement baselines, run experiments. Build research skills while contributing to ongoing work. Often during or right after a Master's/PhD.
Lead independent research projects. First-author publications. Define your own research agenda within the team's focus.
Lead a research area. Multi-paper research programmes. Influence the lab's direction. Significant publication record.
Define research strategy for the lab. Landmark publications. Invited talks, programme committees, industry influence.
Already working in another field? Here is how your background maps.
The most direct path. Focus your thesis on a hot area (LLMs, alignment, reasoning, multimodal). Publish at top venues. Apply for research internships at labs during your PhD.
This is a difficult switch without a graduate degree. Some labs hire strong engineers as research engineers (not research scientists). Consider a part-time Master's or focus on applied research roles that value engineering skills.
Your mathematical training transfers well. Learn ML and deep learning, then apply your quantitative skills to AI research. Physics-to-ML is a well-established pipeline, especially for roles involving optimisation and theory.
