About me

I am a Research Scientist at Meta Reality Labs focused on accelerating machine learning inference on embedded devices. My work encompasses optimizing and deploying diverse ML architectures—including CNNs, RCNNs, RNNs, and transformers—onto resource-constrained hardware platforms. I leverage PyTorch and related frameworks to develop efficient model implementations and quantization techniques for edge deployment. Additionally, I partner with SoC vendors to architect NPU/eNPU specifications that enable more efficient and effective hardware solutions for on-device ML inference.

I received my Ph.D. in Computer Science from the Georgia Institute of Technology, where I specialized in Systems under the advisement of Dr. Taesoo Kim. My thesis, Taming Latency In Data Center Applications, is available here. I also hold an M.S. in Computer Science from Georgia Tech and a B.E. in Computer Science from the University of Madras.

Publications

• ECOTLB: Eventually Consistent TLBs
  Steffen Maass, Mohan Kumar, Taesoo Kim, Tushar Krishna, and Abhishek Bhattacharjee.
  ACM TACO’20.

• SOLROS: A Data-Centric Operating System Architecture for Heterogeneous Computing
  Changwoo Min, Woon-Hak Kang, Mohan Kumar, Sanidhya Kashyap, Steffen Maass, Heeseung Jo, and Taesoo Kim.
  EuroSys’18, Porto, Portugal. (acceptance rate of 16.4%)

• LATR: Lazy Translation Coherence
  Mohan Kumar, Steffen Maass, Sanidhya Kashyap, Jan Vesely, Zi Yan, Taesoo Kim, Abhishek Bhattacharjee, and Tushar Krishna.
  ASPLOS’18, Williamsburg, VA, USA. (acceptance rate of 17.6%)

• Mosaic: Processing a Trillion-Edge Graph on a Single Commodity Machine
  Steffen Maass, Changwoo Min, Sanidhya Kashyap, Woonhak Kang, Mohan Kumar, and Taesoo Kim.
  EuroSys’17, Belgrade, Serbia. (acceptance rate of 20.5%) 🏆 Best student paper

• S-NFV: Securing NFV states by using SGX
  Ming-Wei Shih, Mohan Kumar, Taesoo Kim, and Ada Gavrilovska.
  SDN-NFV Security’16, New Orleans, LA, USA. 🏆 Best Paper and presented in NFV World Congress’16

• TCP Ordo: The cost of ordered processing in TCP Servers
  Mohan Kumar and Ada Gavrilovska.
  INFOCOM’16, San Francisco, CA, USA. (acceptance rate of 18.25%)

Posters

• mKPAC: Kernel Packet Processing for Manycore Systems
  Ramneek, Mohan Kumar, Taesoo Kim, and Sungin Jung.
  Middleware’18, Rennes, France.

• Network Function Fault Isolation in a Single Address Space
  Mohan Kumar, Steffen Maass, and Taesoo Kim
  NSDI’17 Poster, Boston, MA, USA.

• DistCoz: Tell Me What to Optimize in My Distributed Application
  Steffen Maass, Mohan Kumar, and Taesoo Kim
  NSDI’17 Poster, Boston, MA, USA.

• VNFStore: NFV State Externalizing Framework
  Mohan Kumar and Ada Gavrilovska.
  Diversity Workshop at SOSP’15, Monterey, CA.