Description

About IBM:

IBM Research is the innovation and growth engine of IBM and the largest industrial research organization globally, with 12 labs on 6 continents. It produces more breakthroughs—over 9 patents every day—than any other organization. IBM employs over 3,200 researchers worldwide.

IBM Research India (IRL) leads industrial research in India, shaping the future of computing across AI, Hybrid Cloud, and Quantum Computing. IRL has a strong legacy of groundbreaking innovation across computer science disciplines, with work spanning from academic collaborations to enterprise-grade software development. Current focus areas include foundation models, optimized runtime stacks, multi-accelerator optimization, agentic workflows, and applications across language, code, time series, IT automation, and geospatial data.

Role and Responsibilities:

The Research Engineer role at IBM India Research Lab is dynamic, challenging, and highly innovative. The responsibilities include:

• Working on optimized runtime stacks for foundation model workloads, including fine-tuning, inference serving, and large-scale data engineering.
• Focusing on multi-stage tuning, reinforcement learning, inference-time computation, and data preparation for complex AI systems.
• Optimizing models for multiple accelerators, including IBM’s AIU accelerator, leveraging compiler optimizations, specialized kernels, libraries, and tools.
• Developing use cases that effectively leverage IBM’s infrastructure and models to deliver business value.
• Engaging in pre-training language and multi-modal foundation models using large-scale distributed training, model alignment, specialized data pipelines, and human data collection frameworks.

Key Technical Skills:

Machine Learning, big data, Pytorch FSDP & HuggingFace, Python, web development

Requirements:

• Bachelor’s degree in Computer Science, AI, or related fields (Master’s preferred)
• 0-8 years of experience in modern ML techniques such as model architectures, data processing, fine-tuning, reinforcement learning, distributed training, and inference optimization
• Experience with big data platforms like Ray and Spark
• Experience with PyTorch FSDP and HuggingFace libraries
• Programming experience in Python or web technologies
• Growth mindset and pragmatic attitude
• Preferred: peer-reviewed research in top ML or systems conferences, experience with CUDA, Triton kernels, GPU scheduling, memory management, and open-source community contributions

About IBM’s Culture:

IBM fosters an environment where employees are encouraged to be courageous, experiment, and innovate. IBMers are growth-minded, curious, and collaborative, driven to deliver exceptional results. IBM is committed to diversity, inclusion, and equal opportunities for all.

About IBM as an Organization:

IBM has been restlessly reinventing itself since 1911 and is a leading force in technologies like AI, quantum computing, and blockchain. IBM helps clients globally to improve businesses and society, and prides itself on being a responsible technology innovator.

Other Benefits and Information:

• No travel required for this role
• General daytime shift
• Not a commissionable or sales incentive-based role
• IBM is an equal-opportunity employer committed to fair employment practices