Research Experiences:
- [ Sep’15 – Present ] : Research Assistant – Distributed Systems (DPRG)
- Scalable Distributed Strong Failure Detector: Developing a scalable and distributed strong failure detector. Since existing scalable failure detectors (e.g. SWIM) are weak, and on the other hand, strong failure detectors (e.g. Virtual Synchrony) are not scalable, my research aims to develop a distributed failure detector to support the both.
- Microsoft Service Fabric: Analyzed and measured the performance of “Microsoft Service Fabric”- a distributed framework for building microservice-based applications (in collaboration with Microsoft).
- CAT – A New Impossibility Theorem: Proposed a new impossibility theorem called
“CAT” that would suit for transactional distributed database/storage systems
(NewSQL), where NewSQL systems need to consider the impossibility theorem in
terms of Contention, Abort Rate, and Throughput.
- [ Jan’14 – Aug ’15] : Research Assistant – Wireless Networking
- An Opportunistic MAC protocol: Developed an efficient wireless MAC protocol that opportunistically enhances throughput and reduces the backoff time.
Work Experiences:
- [ May’17 – Aug’17 ] : Intern – Microsoft Azure.
- Linearizability Checker for the “Reliable Collection”: Designed a failure resilient microservice-based system that deploys and initiates a database, performs operations on it, injects failure and at the same time generates log. The log is later fed to “Jepsen”, a well-known linearizability checker to check for any inconsistency.
- [ May’16 – Aug’16 ] : Intern – Microsoft Azure.
- Benchmarking Tool: Developed a benchmarking tool using the microservice-based architecture, that is also inspired by the well-known YCSB benchmarking tool, to measure the performance of the Reliable Collection compared to MongoDB and Cassandra. Microsoft also uses this tool for Regression testing
- [ May’14 – Aug’14 ] : Intern – Huawei Technologies.
- A Lightweight “emergency message” Broadcasting Technique: Developed this
technique to spread emergency messages as fast as possible while reducing the
Broadcast Storm in the Vehicular Ad-Hoc Network. I used Vehicle-to-Vehicle communication where a received emergency message was further relayed based on the
received SNR level. NS3 was used to measure the effectiveness of the approach.
- [ Mar’11 – Jul’13 ] : FULL TIME Research Eng. – Samsung R&D Institute Bangladesh.
- Worked on Processor Specific algorithm optimization.
- Implemented and optimized Image Effect Filters for low powered and low memory devices.
- Worked on the development of “Samsung iStudio”, Samsung’s in-house image editing software
Teaching Experiences: