ML Research Engineer | Inference, Model Optimisation and Internships Counted

Cerebras builds wafer scale processors, physically enormous single chips designed to run large AI models far faster than clusters of conventional GPUs, and this role sits on the inference side: making models actually run quickly on that hardware. The work spans model optimisation, evaluating where inference time is going, and the engineering around serving models at speed. What makes this posting worth flagging is its experience line. It asks for 1 to 3 years in software engineering or machine learning, and then states in the same breath that internships count. That is the sentence students look for and almost never find, and it is coming from a frontier AI hardware company rather than a services firm. If you have done ML coursework, a research project or an internship and can write real code, you are inside the stated band rather than hoping to be considered.

Required
- 1 to 3 years of experience in software engineering or machine learning in a similar capacity. The posting adds, in the same sentence, that internships count. That is unusual and it is why this role leads the entry level group today.
- Machine learning and inference work, with the optimisation focus described in the posting.
- Software engineering fundamentals: this is an engineering seat, not a pure research seat.

The day to day
Inference focused machine learning engineering: getting models to run efficiently on Cerebras hardware, working on model optimisation and the serving path, and evaluating performance. Cerebras designs wafer scale chips, so the work sits unusually close to the hardware for an ML role, and the neighbouring openings on the same board (ML systems performance, post silicon bring up, physical design) show how deep that stack goes.

Location and working style
The posting lists the India office, and Cerebras states elsewhere on its board that it can work in a hybrid environment. Confirm the exact office and the number of days on site at the first screen, because the listing gives the location as India Office rather than naming a city.

Honest fit guidance
This is the most accessible genuine engineering role on today's list, and the internships clause is the reason. Treat "1 to 3 years" as the real bar rather than a formality, but do not add imaginary requirements on top of it: the company wrote down that project and internship work counts.

Be clear eyed about the domain. This is inference and systems work at an AI chip company, so comfort with performance, profiling and low level thinking will matter more than familiarity with the latest model architectures. If your ML experience is entirely notebooks and fine tuning APIs, expect the interview to push toward engineering fundamentals.

Cerebras has five roles on this edition across very different levels, from this one at 1 to 3 years up to a 10+ year full stack ML lead. They are separate teams, not one requisition posted five times, and applying to the level that matches you is worth more than applying to all of them.