Introduction
At IBM, work is more than a job – it’s a calling: To build. To design. To code. To consult. To think along with clients and sell. To make markets. To invent. To collaborate. Not just to do something better, but to attempt things you’ve never thought possible. Are you ready to lead in this new era of technology and solve some of the world’s most challenging problems? If so, lets talk.

Your Role and Responsibilities
The Data and Model Factory team in Cambridge is looking for a talented and highly motivated engineer to help advance our effort on creating the most efficient foundation models.
The candidate will be responsible for conducting cutting-edge research on natural language processing, developing prototype solutions to real-world problems, working closely with top-notch MIT faculty, students, and IBM scientists in a flexible and fun environment.

Required Technical and Professional Expertise
We seek an engineer to refine and standardize our large language model training procedures. This role is crucial for enhancing our existing framework, encompassing a broad spectrum of tasks from dataset preprocessing to large-scale distributed training techniques such as fine-tuning, contrastive fine-tuning, RLHF, RLAIF, EvolInstruct, AutoGen, etc. The candidate will be instrumental in developing tools and scripts that streamline the entire alignment pipeline, ensuring efficiency and consistency. This position addresses our current need for expertise in model training, experimentation, standardizing reporting results, and integrating robust DevOps and MLOps practices.

The engineer will be responsible for supporting all tooling and software engineering efforts to standardize and optimize the alignment pipeline. They should be able to interpret machine learning research and translate it into reliable, maintainable software.

Additionally, the role involves developing specialized pipelines for various tasks, such as building RLHF pipelines with unit test-based rewards for low-resource languages and creating effective LLM-generated data pipelines. The ideal candidate will have a passion for programming languages, enabling them to tailor data generation and training pipelines to specific languages, and if applicable, leverage webapp development skills for creating frameworks to collect human data and deploy models in user-centric platforms.

Preferred Technical and Professional Expertise
Strong programming skills.
Experience with machine learning tools and frameworks such as TensorFlow, PyTorch etc.
Experience with large language models.