Jun 2022 – Oct 2024 · 2 yrs 4 mos

Jan 2022 – Jun 2022 · 5 mos

Nov 2019 – Dec 2021 · 2 yrs 1 mo

Aug 2015 – Nov 2019 · 4 yrs 3 mos

May 2013 – May 2013 · 0 mo · Brussels Metropolitan Area

• External evaluator for the Clean Sky Public Private Partnership between the European Commission and the Aeronautical Industry, EU FP7

Sep 2012 – Mar 2013 · 6 mos · Bangaon, West Bengal, India

• Performed field failure data analysis for one of the Infosys flagship SaaS platforms.Identified reliability bottlenecks. Communicated the findings of the study to the CTO, as well as to the development and verification teams. Identification of reliability bottlenecks in platform codebase. Findings resulted in a 72% improvement of the end to end reliability of the main processing components
• o Created a data-driven stochastic hazard model. The model allowed the improvement of virtual machine migration and customer handling policies. The model and failure analysis tool formed the basis of an internal diagnosis system used at Infosys.
• o Analyzed ~ 300,000 production SaaS transactions discovering about 300 SLA violations. Performed fault injection experiments to diagnose the causes of SLA violations. Injections were performed on a testbed replicating the production environment. Results enabled developers to identify fixes that will be part of the next release of the SaaS platform software.

Jan 2012 – Aug 2015 · 3 yrs 7 mos · Urbana-Champaign Area

Oct 2009 – Dec 2011 · 2 yrs 2 mos

• Developed a technique and tools for the assessment of time-based coalescence heuristics for supercomputers error log analysis. The technique enabled the generation of synthetic error logs using stochastic models interfacing with system-level logging APIs. The models were developed starting from the analysis of publicly available supercomputer failure data. The approach allowed the creation of a new error coalescence heuristic capable of delivering accurate results at the scale of petascale systems and above.
• Implemented a predecessor of modern digital twins for mission-critical systems by creating a framework for the simulation-in-the-loop assessment of wireless sensor networks for critical infrastructure monitoring. The framework was able to generate performance and dependability models on the fly from real data.

Oct 2006 – Oct 2009 · 3 yrs

• Resiliency of Wireless Sensor Network for critical infrastructures
• Modeling of large scale critical systems
• Event log analysis of large scale computing systems

Oct 2007 – May 2009 · 1 yr 7 mos

• Security and Resiliency of Cloud computing
• Data analytics, event log analysis, checkpointing, workload analysis and modeling of supercomputers
• Design of Dependable and Secure Wireless Sensor Networks
• Data Analysis and dependability of medical devices

Sep 2001 – Oct 2006 · 5 yrs 1 mo