Feb 2024 – Present · 2 yrs 1 mo

• Authored and presented a paper on AI-driven datacenter failure prediction at CUG2024, engaging with global customers and industry experts in Perth, Australia.
• Authored a paper on the sustainability impact of AIOps in datacenters, shortlisted for presentation and publication at the Sustainable Supercomputing Workshop, part of the SC24 conference.
• Authored and presented multiple papers at HPE internal technical conferences.
• Led a team to build a RAG based chat interface for product documentation, improving information accessibility.
• Designed and implemented a machine learning model to predict HPC job failures at the submission phase, reducing wasted compute cycles.
• Achieved 3× improvement in model inference throughput using hardware accelerators. (TensorRT and OpenVINO).
• Led development of AI-powered HPC diagnostics solutions enhancing customer reliability and support experiences.
• Built a highly scalable real-time log anomaly detection system for monitoring datacenter system logs using AI/ML.
• Developing an intelligent alert correlation engine that clusters related datacenter alerts, improving incident response.
• Leading design and development of an advanced multi-agent log analytics platform for autonomous system monitoring.
• Enhancing and maintaining AIOps pipelines deployed in customer datacenters, to ensure reliability and continuous improvements.
• Developing an agentic AI-powered copilot system for next-generation HPC system management software.

Feb 2022 – Feb 2024 · 2 yrs

• Partnered with customers to enhance and optimize machine learning pipelines deployed in production environments.
• Designed and deployed an AI-powered predictive tool to forecast datacenter failures, enabling proactive mitigation.
• Built a POC anomaly detection system leveraging system log analytics to identify abnormal behaviors.
• Contributed to transforming AIOps into a cloud-native service, improving scalability and accessibility.
• Developed a machine learning-driven forecasting tool for cluster diagnostic results, improving operational efficiency.
• Authored and presented multiple papers at HPE internal technical conferences.

Aug 2019 – Jan 2022 · 2 yrs 5 mos

• Developed an end-to-end AI system to detect anomalies across thousands of datacenter sensors with high efficiency.
• Validated the solution on customer-provided historical data, achieving:
• o Early detection of anomalies up to 5 minutes before customer-reported incidents.
• o Identification of 50% more anomalies than existing systems.
• o Demonstrated potential to prevent 40% of high-priority incidents that could escalate into critical events.
• Integrated AIOps with HPE and Cray’s cluster management platforms, enabling proactive monitoring for some of the world’s fastest supercomputers.

Jan 2019 – Jul 2019 · 6 mos

• • Developed statistical and deep learning models for anomaly detection on large-scale datacenter sensor telemetry datasets (tens of millions of records) provided by our customers.