Tushar Kumar Barman
Software Development Engineer
Building cool thingsAbout
- GPU Software Development Engineer at Intel Corporation, working on heterogeneous computing, IDE toolchains, and infrastructure observability for next-gen GPU platforms.
- Published researcher at IEEE Cluster 2023 — analyzed how different BeeGFS configurations affect I/O performance in HPC environments, revealing that default allocation strategies lead to imbalanced workload distribution.
- Passionate about building intelligent systems — from AI-powered CI/CD analysis agents to real-time telemetry pipelines across 300+ distributed systems.
- Master's in Computer Science from BITS Pilani (GPA: 8.95, Top 10 in class).
Connect
Skills
Languages
Tools & Frameworks
DevOps & Cloud
Experience
- Integrated Intel Compiler IDE extensions for Visual Studio 2026, providing early-access support for next-gen heterogeneous computing
- Managed end-to-end release lifecycle of Intel oneAPI Compiler extensions, maintaining 100% backward compatibility
- Developed a Python-based AI agent to parse massive CI/CD test logs and generate JIRA tickets with root cause details
- Built an observability pipeline for hybrid cloud environments, centralizing real-time telemetry for 300+ distributed CI systems
- Created Grafana dashboards adopted by 4+ teams, improving infrastructure utilization by 20%
- Developed backend using NestJS and Elasticsearch with server-side aggregations and paginated REST APIs
- Built Python data collectors using GitHub Actions to normalize API streams into unified JSON schema with Zod validation
- Developed a Python-based ETL pipeline using multi-threading to ingest telemetry from OneCloud and GTAX APIs
- Maintained and improved production Grafana dashboards for 4+ engineering stakeholders
- Created scripts for classifying 1M+ log lines into training sets for predictive failure detection
Research
Analyzed how different BeeGFS parallel file system configurations affect I/O performance in high-performance computing environments. Found that default allocation strategies and striping configurations lead to imbalanced workload distribution, negatively impacting I/O throughput.
Projects
Python-based AI agent that parses massive CI/CD test logs, extracts critical failure data, and uses OpenAI API to generate concise root-cause reports and auto-create JIRA tickets.
Real-time telemetry pipeline for hybrid cloud environments, centralizing monitoring across 300+ distributed CI systems. Built dashboards adopted by 4+ teams, improving infra utilization by 20%.
Cloud-based Intrusion Detection System built with Docker and the ELK Stack on AWS, enabling real-time threat detection and security event visualization.
Published at IEEE Cluster 2023 — analyzed the impact of BeeGFS configuration on HPC I/O performance using ELK-based monitoring across distributed file systems.
Education
"I was not born with a whole lot of natural talent... but I work hard and I never give up."