Rajni Pawar
About
I am a PhD student in Computer Science at Illinois Institute of Technology, passionate about advancing scientific computing through innovative I/O optimization and artificial intelligence integration.
Currently conducting research at Gnosis Research Center under Dr. Xian-He Sun, focusing on accelerating data-intensive scientific workflows. My work involves developing tokenization-integrated I/O systems, fine-tuning large language models for deployment automation, and building AI tools for the National Data Platform.
My research interests span high-performance computing, distributed systems, AI/ML for scientific applications, and developing efficient data management solutions that bridge the gap between storage systems and modern AI workloads.
Research Experience
Graduate Research Assistant
Gnosis Research Center, Illinois Institute of Technology
Jan 2025 - Present | Advisor: Dr. Xian-He Sun
Enhanced Scientific Exploration with AI
Contributing to IOWarp, a data management platform that accelerates data-intensive workflows. Fine-tuned LLMs (granite, phi-4-mini-reasoning) to support Jarvis, a deployment tool under IOWarp. Leveraged Unsloth, LoRA, and GPUs for faster computation achieving 95% accuracy. Built a Model Context Protocol (MCP) server integrating AI tools with the National Data Platform, accelerating AI-driven research workflows by 40%.
An Active Tokenizing I/O Stack
Developed tokenization-integrated I/O stack storing ready-to-use tokens for LLMs and RAG workloads. Accelerated AI pipelines by 57.1% eliminating preprocessing overhead. Validated scalability on HPC with MPI and SLURM.
Networking Investigation and Deployment Strategies
Created unified network library exploring ZeroMQ, Libfabric, and Thallium achieving 50% faster data transfer. Developed deployment script with dynamic hostfile generation and nodeAffinity constraints. Orchestrated containerized scientific workloads across HPC-deployed Kubernetes clusters.
Publications
Research Poster Presentation
"Scaling Data Tokenization for AI Systems"
SSDBM 2025 Conference
View Full Poster PDFM. Kolhekar, S. Kurle, R. Pawar, J. S. Kumar and R. Verma
"Stylized Devanagari Script License Plate Conversion to Standard Script using Deep Learning"
ICAST 2023, pp. 143-148, doi: 10.1109/ICAST59062.2023.10454944
Work Experience
System Engineer
Tata Consultancy Services
June 2021 - July 2024 | Mumbai, India
- Implemented high-performance computing workflows using C++ and CUDA, optimizing for GPU-accelerated processing with SLURM and LSF, reducing computational time by 50%
- Engineered enterprise-grade website with 50+ scalable AEM components, collaborating with UI/UX designers using HTML, CSS, and JavaScript, resulting in 30% increase in user metrics
- Integrated AEM components using Java, leveraging Jenkins CI/CD for automated deployments, supporting 5M+ users
Other Projects
SimpleChat - Distributed Messaging Application
Developed a multi-node real-time messaging system using C++ and Qt6 with TCP socket programming, implementing ring network topology for automated message routing between interconnected nodes with modern GUI and conversation management.
DevPlate: License Plate Recognition System
Engineered a recognition system for stylized Devanagari script license plates, converting them to standard scripts using KNN algorithm, OpenCV, and EasyOCR with 85% accuracy to aid law enforcement authorities.
Education
Illinois Institute of Technology
Doctorate of Philosophy in Computer Science
Aug 2024 - Present | Chicago, IL
GPA: 3.83/4.0
Relevant Coursework: Advanced OS, Algorithms, ML, Computer Networks
University of Mumbai
Bachelor of Engineering - Electronics and Telecommunication Engineering
June 2021 | Mumbai, India
CGPA: 9.07/10
Technical Skills
Programming
Systems
AI/ML
Cloud/DevOps
Certifications & Awards
AWS Certified Solutions Architect - Associate
Microsoft Certified: Azure Fundamentals
Won TCS Ideathon 2021-2022 for Smart Package Manager