**Rajath Shashidhara** Graduate Student (MS Computer Science) The University of Texas at Austin rajaths@cs.utexas.edu ![ ](rajath.jpg width="28%" border="2") I am a graduate student in the [Department of Computer Science](https://www.cs.utexas.edu/) at The [University of Texas at Austin](https://www.utexas.edu). Previously, I was a member of the 5G Communication R&D Team at Samsung Research, Bangalore. I have a keen interest in research and development of scalable, efficient and resilient computing systems. Presently, I am working on building high-performance TCP stacks accelerated by SmartNICs advised by Prof. [Simon Peter](http://www.cs.utexas.edu/~simon/). You can find my resume [here](cv.pdf).

[](https://github.com/rajathshashidhara)   [](https://scholar.google.com/citations?user=47vhbO8AAAAJ&hl=en)   [](https://www.linkedin.com/in/rajath-s)   [](https://twitter.com/rajaths_)

Latest ------------------------------------------------------------------------------- - **IEEE Globecom 2020** A Reinforcement Learning Framework for QoS-Driven Radio Resource Scheduler - work done at Samsung Research. Find abstract in Section [Research] - [Lambda-KV](lKV.pdf): Distributed Key-Value Store with colocated Serverless Compute Education =============================================================================== M.S.     2019 - 2021 : _Computer Science_ :
*The University of Texas at Austin* (UT Austin) :
Thesis (tentative): _Offloading TCP stacks to SmartNICs_ :
Advisor: [Dr. Simon Peter](http://www.cs.utexas.edu/~simon/) :
CGPA: _4.0_ :
Courses: Advanced Operating Systems, Datacenters, Virtualization, Distributed Computing, Graduate Algorithms :
Teaching Assistant for [Cloud Computing (S20)](https://github.com/devdattakulkarni/CloudComputing/) under [Prof. Devdatta Kulkarni](https://www.linkedin.com/in/devdatta-kulkarni-192ab46/) M.Sc.     2012 - 2017 : _Physics_ :
*Birla Institute of Technology and Science (BITS), Pilani* :
Thesis: [_Driven Aubry-André-Harper Models_](thesis.pdf) :
Advisor: [Dr. Tapomoy Guha Sarkar](https://scholar.google.com/citations?user=FLFD3-gAAAAJ&hl=en) :
Adjudged *Best Student of Batch 2017* for outstanding academic and research track record :
CGPA: _9.01/10_ B.E.     2012 - 2017 : _Computer Science_ :
*Birla Institute of Technology and Science (BITS), Pilani* :
CGPA: _9.01/10_ (Major GPA: _9.33/10_) :
Received 80% scholarship every semester (≤5% out of 800 students) Experience =============================================================================== Summer 2020 : *Confluent*, Mountain View CA :
_Software Engineering Intern_, Cluster Management :
Manager: [Lokesh Shekar](https://www.linkedin.com/in/lokesh-shekar-a021b822/) :
+ Development of Kubernetes Operator for automating deployment life-cycle of complex stateful Kafka clusters :
+ Designed *safe & seamless* live migration of Kafka clusters between different operators. No service disruption/data loss during migration! June 2017 - August 2019 : *Samsung Research*, Bangalore INDIA :
_Senior Software Engineer (Research)_, Communication R&D :
Manager: [Anshuman Nigam](https://www.linkedin.com/in/anshuman-nigam-3485661/) and [Dr. Dojun Byun](https://www.linkedin.com/in/do-j-byun-3911425b/) :
+ Developed the fast data-plane radio access network stack (PDCP, RLC, MAC) for the world’s first Pre-5G mobile user equipment. Technical support for the 5G demo at _PyeongChang Winter Olympics (Korea, 2018)_ :
+ Research on _parallelization, memory management & flow control_ to improve throughput and reduce memory footprint of datapath of 5G NR Distributed Unit (commercialized in USA & Korea) on hardware-accelerated platforms (Marvell/Cavium) and DPDK :
+ Developed _Reinforcement Learning based QoS-aware Radio-Resource Scheduling_ algorithms: multi-objective optimization in stochastic input-driven environments _(IEEE Globecom 2020)_ :
+ Presented with _Samsung Technical Excellence Award_ for no critical S/W bugs in bare-metal real-time 5G Network Stack Spring 2017 : *Symantec Inc*, Bangalore INDIA :
_Software Engineering Intern_, Website Security Development Team :
+ Designed a proof-of-concept microservices based cloud-ready web application to automate the purchase, delivery & installation of SSL certificates for webservices hosted on Amazon AWS Summer 2016 : *Microsoft R&D*, Hyderabad INDIA :
_Software Engineering Intern_ :
+ Integrated Azure AD cloud authentication/authorization service into ASP.NET Core based web applications Summer 2015 : [*Graduate Institute of Astronomy*](http://www.astro.ncu.edu.tw/index_e.php) - National Central University, Zhongli, Taiwan :
Advisor: [Dr. Ko, Chung-Ming](http://www.astro.ncu.edu.tw/people/faculty_bio_e.php?id=3) :
_Summer Research Intern_ :
Studied _Gravitational Lensing in Elliptical Galaxies_ :
+ Analytically [derived](report_gl.pdf) the gravitational lensing equation for elliptical galaxies. :
+ Developed a [distributed recursive sub-gridding algorithm](https://github.com/rajathshashidhara/gravitational-lensing-elliptical-galaxies) to find the solution on a cluster and simulate the lensing. Summer 2014 : [*Bhaskaracharya Institute of Space Applications and Geoinformatics*](https://bisag.gujarat.gov.in/), Gandhinagar, India :
_Summer Research Intern_ :
+ Surveyed existing [literature](report_stitch.pdf) on algorithms to [stitch](https://github.com/rajathshashidhara/GeoImageStitching/) large satellite images into a mosaic. :
+ Evaluated performance of OpenCV implementations of [SIFT & SURF](slides_stitch.pdf) algorithms on large datasets of satellite images Summer 2013 : [*Google Summer of Code 2013*](https://www.google-melange.com/archive/gsoc/2013/orgs/apache/projects/rajaths589.html) :
_Software Development Intern_, Apache Software Foundation - OpenOffice :
+ [Developed](https://github.com/rajathshashidhara/CMISContentProvider) an in-app document version control toolbar which connects to cloud content repository Research =============================================================================== - *Lambda-KV: Distributed Key-Value store with colocated Serverless compute*
[Paper](lKV.pdf)   •   [Code](https://github.com/rajathshashidhara/SimpleDB)
Advisors: [Prof. Vijay Chidambaram](https://www.cs.utexas.edu/~vijay/) and [Prof. Simon Peter](http://www.cs.utexas.edu/~simon/) - Serverless execution framework aggregates compute and storage resources into a single entity that aggressively exploits data locality to significantly improve efficiency. - Lambda-KV improves the overall execution time of serverless compilation workloads by a factor of 1.6x and 5x reduction in time spent in data movement against conventional disaggregated deployments. - *SRoCE: Software RDMA over Commodity Ethernet*
[Paper](sRoCE.pdf)   •   [Code](https://github.com/mani-shailesh/rdma-tas)
Advisors: [Prof. Simon Peter](http://www.cs.utexas.edu/~simon/) and [Prof. Chris Rossbach](http://www.cs.utexas.edu/~rossbach/) - Software-based flexible RDMA verbs implementation using high performance user-space TCP stack. - Achieved 3x single-connection throughput for 1000 bytes RDMA ops as compared to H/W RDMA NICs. - *Improving Connection Scalability in [TAS: TCP Acceleration as an OS Service](https://people.mpi-sws.org/~antoinek/documents/19eurosys_tas.pdf)*
[Paper](tas_scale.pdf)   •   [Code](https://github.com/rajathshashidhara/tas)
Advisor: [Prof. Simon Peter](http://www.cs.utexas.edu/~simon/) - Increased throughput by 10% at 100k connections by improving the performance of flow pacing algorithm. - Overhead of Queue Manager – Fair Queueing (FP) based flow scheduler in TAS increases with the number of connections. In place of it, we use [Carousel](https://www.cc.gatech.edu/~amsmti3/files/carousel-sigcomm17.pdf) to achieve ~1.2x times higher throughput and decrease in CPU utilization of enqueue operation by 50% in comparison with FQ pacing for greater than 16k connections. - *A Reinforcement Learning Framework for QoS-Driven Radio Resource Scheduler* - 5G MAC scheduler that learns an allocation policy to simultaneously optimize multiple, often conflicting QoS objectives. - Approach allows network operators to customize their requirements, by assigning priority values to QoS classes. - Presents a flexible neural-network architecture that can easily adapt to varying number of flows, drastically simplifying training, thus rendering it viable for practical implementation in constrained systems. - Outperforms conventional heuristics such as M-LWDF, EXP-RULE and LOGRULE and is robust to changes in radio environment and traffic patterns. - *Studying Quantum Chaos in Aubry-André electron systems*
[Paper](https://doi.org/10.1103/PhysRevA.94.053612)   •   [Code](https://github.com/rajathshashidhara/DiscreteSchrodinger)   •   [Thesis](thesis.pdf)
Advisor: [Dr. Tapomoy Guha Sarkar](https://scholar.google.com/citations?user=FLFD3-gAAAAJ&hl=en) - Studied phase transitions in Hofstadter’s butterfly under time-varying magnetic field and the relationship between topological invariants and Hall conductivity. - Simulated and computationally evaluated solutions to Schrodinger’s equation for special quantum systems using perturbation methods and computational physics algorithms. - *Distributed Combinatorial Optimization on a Cluster*
[Report](design_bnb.pdf)   •   [Code](https://github.com/rajathshashidhara/Distributed-Branch-and-Bound)
Advisor: [Prof. Sundar Balasubramaniam](https://www.researchgate.net/profile/Sundar_Balasubramaniam) - Designed a distributed algorithm to efficiently perform Branch & Bound search on a commodity cluster. - Developed a load balancing technique based on peer-to-peer diffusion between nodes on toroid communication topology and leftist-heap based work-stealing queues between threads. Publications =============================================================================== - *A Reinforcement Learning Framework for QoS-Driven Radio Resource Scheduler*
Jitender Singh Shekhawat, Rishabh Agrawal, K Gautam Shenoy, *Rajath Shashidhara* @ Samsung Research, Bangalore
To appear in _IEEE Globecom 2020_ - *[Phase transition in an Aubry-André system with a rapidly oscillating magnetic field](https://doi.org/10.1103/PhysRevA.94.053612)*
Tridev Mishra, *Rajath Shashidhara*, Tapomoy Guha Sarkar and Jayendra N. Bandyopadhyay
Physical Review A 94, no. 5 (2016): 053612. - *[Driven Aubry-Andre-Harper Systems](thesis.pdf)*
*Rajath Shashidhara*
Master's Thesis, Birla Institute of Technology and Science, Pilani (May 2017) Projects =============================================================================== - *Implementing Paravirtual Hypervisor in JOS*
Advisor: [Prof. Vijay Chidambaram](https://www.cs.utexas.edu/~vijay/)
[Course Page](https://github.com/vijay03/cs378-f19)   •   [Related Paper](http://www.cs.unc.edu/~porter/pubs/jos-vm.pdf) - Built a basic paravirtual hypervisor using x86 hardware virtualization features to host JOS-on-JOS - *Fast Semantic matching of strings generated by Context Free Grammar*
Advisor: [Prof. Sundar Balasubramaniam](https://www.researchgate.net/profile/Sundar_Balasubramaniam)
[Report](report_cfg.pdf)   •   [Code](https://github.com/rajathshashidhara/SemanticMatchingCFGs) - Designed a language for domain experts to express semantic equivalence based on parse tree structure - Developed a hash function to hash parse trees for fast matching. Experimented on XML DBs - *Persistent storage with C++ STL abstraction* - Implemented templatized out-of-core (secondary storage) data structures (B+ Trees, Vectors) with STL interface - User-space applications simply need to relink with [library](https://github.com/ExternalMemoryDS/external-mem-ds) for persistent structures - Customized buffer caches by passing the kernel, async I/O for high efficiency - Built a proof-of-concept TF-IDF based [Search Engine](https://github.com/rajathshashidhara/InformationRetrievalSystem) using this library that scales beyond primary memory limits (> 100GB) - *IoT enabled Laboratory Environment: Project SmartLAB*
[Blog](https://smartlabbits.wordpress.com/)   •   [Code](https://github.com/smartlab-bits) - Proactive lab monitoring and activity tracking using sensor networks, speech and gesture recognition - Awarded *Prof. I J Nagrath Student Project Fund* by Dept. of Electrical Engineering, BITS Pilani - Runner-up in Seimens Home Automation Challenge (Apogee 2014, BITS Pilani)