GRM is a unique platform where students remotely work on challenging short/long term technical projects defined by IBM mentors.
Students get the opportunity to interact with top IBM technical expertise and learn to use tools and strategies to get skilled and comfortable with gen next technologies and doing it online makes it easier to work from any part of India.
The goal is to explore causal inference in time series and see if foundation models can help in the task.
Integration of profiling into torch.compile workflow with inductor backend, by identifying the following: 1. Which metrics to collect - timing, register access counters, etc 2. Where to insert hooks - Fx graph, Inductor IR, Triton IR, etc 3. Areas
Build notebooks to increase open-source adoption of TSFM models
1. To build an AI-driven evaluation system where AI agents assess and improve the performance of other AI agents. 2. To implement a multi-layer coaching pipeline (Primary Generator → Lightweight Coach → Heavy Coach) for scalable, reliable, self-eval
To produce an embedding model that responds well to user provided instructions
Explore AI-assisted techniques to modernize Assembly programs into high-level language like C. Leverage recent advances in Large Language Models (LLMs) to generate human-readable, maintainable code from legacy systems. Evaluate how intermediate rep
This project proposes the development of AdaptiveGraphMAS, a reinforcement learning-driven framework for autonomous construction and optimization of multi-agent system (MAS) topologies. Inspired by the MasHost paradigm, this research aims to advance
Build Newer architectures, benchmarks, and leaderboards in the area of Time-Series Foundation Models. Publish the work in top-tier conferences.
Project Overview Network operation teams often face thousands of cryptic traps and syslogs per day. This Watsonx.ai-based solution will use LLMs enhanced with Retrieval-Augmented Generation (RAG) to translate raw logs into plain English, classify th