Charith Mendis

Office: 4118, Siebel Center for Computer Science
201 N. Goodwin Ave., Urbana IL 61801
Email: charithm (at) illinois.edu

New! Check out our new course on CS521: ML and compilers

Bio

Charith Mendis is an Assistant Professor in the Siebel School of Computing and Data Science at the University of Illinois at Urbana-Champaign. His broad research interests are at the intersection of compilers, program optimization and machine learning. He received his Ph.D. and Master’s from the Massachusetts Institute of Technology and his B.Sc. from the University of Moratuwa. He is the recipient of a DARPA Young Faculty Award, an NSF CAREER Award, a Google ML and Systems Junior Faculty Award, Outstanding Advisor Award at UIUC, the William A. Martin Outstanding Master’s Thesis Award at MIT and the University Gold Medal for his B.Sc. He has won numerous paper awards including a Distinguished Paper Award at POPL, a Best Student Paper Award at the IEEE BigData conference, an honorable mention for the Best Artifact Award at SIGMOD, a Best Paper Award at ML for Systems workshop at ISCA and an IEEE Top Picks Honorable Mention.

Research

It has been challenging to keep compilers up to date with workload and hardware changes. This problem is exacerbated in fast-evolving fields such as machine learning (ML). Traditional means of constructing compilers are too slow to adapt to these changes. My research agenda directly addresses these problems and brings agility and evolvability to compiler construction. Specifically, my group’s (ADAPT lab) work revolves around building the next-generation programming abstractions and compiler construction methodologies that can handle the diversity, complexity, and rapid evolution of both workloads and hardware, while preserving strict guarantees. Our research is centered around three main thrusts.

Automated and Parameterized Compiler Construction Methodologies
- Formalism-driven transformation space construction: Generating compiler components for accelerators (Accelerator Compiler Toolkit) [MICRO’25, POPL’25] and for commodity hardware [PLDI’25, ASPLOS’24, ASPLOS’21]
- Data-driven cost models and optimization strategies: ML-based cost models for commodity hardware [ICML’19, IISWC’22, MICRO’20], accelerators [MLSys’21, ICML’25] and sparse workloads [ASPLOS’23, CGO’26], learned optimizations [NeurIPS’19]
Model-agnostic Compiler Verification and Optimizations for ML
- Tensor Compiler Optimizations: Virtual Tensor Compilation [OSDI’26]
- Tensor Compiler Verification: TensorRight [POPL’25]
- Abstractions and compilation for neural network verifiers: ConstraintFlow [SAS’24, OOPSLA’25]
Model-aware Abstractions and Compiler Optimizations for ML and data science
- Data science [SIGMOD’24, SIGMOD’25]
- Graph Neural Networks: for static [OOPSLA’25, CGO’26, NeurIPS’23] and temporal [ASPLOS’24, PPoPP’23, SIGIR’24] GNNs
- Sparse Neural Networks: for sparse convolution networks [MLSys’23] and sparse transformers [OOPSLA’25, EuroSys’26]

Research Opportunities

ADAPT lab has opportunities for undergraduate and graduate students. If you are a student who is passionate about building high performance ML optimization or compilation techniques or is passionate about automated compiler construction techniques using either machine learning or formal methods, please be in touch with me.

Click to read instructions before contacting me.

Undergraduates or Master's students at UIUC: Please include a CV, an up-to-date transcript, and your programming experience. The subject line should contain the word "prospective undergraduate/master's researcher" to indicate these instructions have been read. Please note that, since we are doing systems work at least 2 semesters worth of commitment is required to get fruitful results (e.g. a publication) out of your experience. Also, I will not agree to supervise any final year Master's students.

Prospective PhD students: First, you must apply and gain admission to the graduate program in Computer Science at UIUC. You should mention me as a potential advisor in your application as well as in your personal statement. Once you apply, you can optionally send me an email with subject "prospective PhD student". If you are admitted, then I'm happy to discuss supervision.

PhD Students

Damitha Lenadora
Stefanos Baziotis
Avaljot Singh (co-advised with Gagandeep Singh)
Devansh Jain
Jai Arora
Chamika Sudusinghe
Muyan Hu (co-advised with Vikram Adve)
Heng Zhong
Zirui Zhou

Selected Publications (all)

POPL

TensorRight: Automated Verification of Tensor Graph Rewrites

Jai Arora , Sirui Lu , Devansh Jain , Tianfan Xu , Farzin Houshmand , Phitchaya Mangpo Phothilimthana , Mohsen Lesani , Praveen Narayanan , Karthik Srinivasa Murthy , Rastislav Bodik , Amit Sabne , and Charith Mendis

In 52nd ACM SIGPLAN Symposium on Principles of Programming Languages , Jan 2025

Distinguished Paper Award

First verification work on production-level tensor compilers used in ML

PDF arXiv Bib Code

@inproceedings{tensorright-popl25,
  author = {Arora, Jai and Lu, Sirui and Jain, Devansh and Xu, Tianfan and Houshmand, Farzin and Phothilimthana, Phitchaya Mangpo and Lesani, Mohsen and Narayanan, Praveen and Murthy, Karthik Srinivasa and Bodik, Rastislav and Sabne, Amit and Mendis, Charith},
  title = {TensorRight: Automated Verification of Tensor Graph Rewrites},
  booktitle = {52nd ACM SIGPLAN Symposium on Principles of Programming Languages},
  year = {2025},
  award = {Distinguished Paper Award},
  impact = {First verification work on production-level tensor compilers used in ML},
  issue_date = {January 2025},
  publisher = {Association for Computing Machinery},
  address = {New York, NY, USA},
  volume = {9},
  number = {POPL},
  url = {https://doi.org/10.1145/3704865},
  doi = {10.1145/3704865},
  journal = {Proc. ACM Program. Lang.},
  month = jan,
  articleno = {29},
  numpages = {32},
  keywords = {Denotational Semantics, Tensor Compilers, Unbounded Verification},
  ae_functional = true,
  ae_available = true
}

MICRO

TAIDL: Tensor Accelerator ISA Definition Language with Auto-generation of Scalable Test Oracles

Devansh Jain , Marco Frigo , Jai Arora , Akash Pardeshi , Zhihao Wang , Krut Patel , and Charith Mendis

In Proceedings of the 58th IEEE/ACM International Symposium on Microarchitecture , Jan 2025

First ISA definition language for accelerators

Used by Amazon for their accelerator offerings

PDF Bib Code

@inproceedings{taidl-micro2025,
  title = {{TAIDL:} Tensor Accelerator ISA Definition Language with Auto-generation of Scalable Test Oracles},
  author = {Jain, Devansh and Frigo, Marco and Arora, Jai and Pardeshi, Akash and Wang, Zhihao and Patel, Krut and Mendis, Charith},
  year = {2025},
  isbn = {9798400715730},
  publisher = {Association for Computing Machinery},
  impact = {First ISA definition language for accelerators, Used by Amazon for their accelerator offerings},
  address = {New York, NY, USA},
  url = {https://doi.org/10.1145/3725843.3756075},
  booktitle = {Proceedings of the 58th IEEE/ACM International Symposium on Microarchitecture},
  pages = {1316–1333},
  numpages = {18},
  ae_functional = true,
  ae_available = true,
  ae_reproduced = true
}

OSDI

VTC: DNN Compilation with Virtual Tensors for Data Movement Elimination

Muyan Hu , Ahan Gupta , Jiachen Yuan , Vima Gupta , Taeksang Kim , Xin Xu , Janardhan Kulkarni , Ofer Dekel , Vikram Adve , and Charith Mendis

In 20th USENIX Symposium on Operating Systems Design and Implementation (OSDI) (to appear) , Jan 2026

Novel optimization that goes beyond operator fusion

PDF

SIGMOD

Dias: Dynamic Rewriting of Pandas Code

Stefanos Baziotis , Daniel Kang , and Charith Mendis

In Proc. ACM Manag. Data (SIGMOD) , Mar 2024

Honorable Mention for the Best Artifact Award

PDF arXiv Bib Code

@inproceedings{dias-sigmod2024,
  author = {Baziotis, Stefanos and Kang, Daniel and Mendis, Charith},
  booktitle = {Proc. ACM Manag. Data (SIGMOD)},
  title = {Dias: Dynamic Rewriting of Pandas Code},
  year = {2024},
  issue_date = {February 2024},
  publisher = {Association for Computing Machinery},
  address = {New York, NY, USA},
  volume = {2},
  number = {1},
  url = {https://doi.org/10.1145/3639313},
  doi = {10.1145/3639313},
  journal = {Proc. ACM Manag. Data (SIGMOD)},
  month = mar,
  articleno = {58},
  numpages = {27},
  keywords = {cross-representation, dynamic, pandas, rewriting},
  award = {Honorable Mention for the Best Artifact Award},
  ae_functional = true,
  ae_available = true,
  ae_reproduced = true
}

ASPLOS

TGLite: A Lightweight Programming Framework for Continuous-Time Temporal Graph Neural Networks

Yufeng Wang , and Charith Mendis

In Proceedings of the 29th ACM International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS) , Mar 2024

PDF Bib Code

@inproceedings{tglite-asplos2024,
  author = {Wang, Yufeng and Mendis, Charith},
  title = {TGLite: A Lightweight Programming Framework for Continuous-Time Temporal Graph Neural Networks},
  booktitle = {Proceedings of the 29th {ACM} International Conference on Architectural
                 Support for Programming Languages and Operating Systems (ASPLOS)},
  publisher = {Association for Computing Machinery},
  year = {2024},
  url = {https://doi.org/10.1145/3620665.3640414},
  doi = {10.1145/3620665.3640414},
  ae_functional = true,
  ae_available = true,
  ae_reproduced = true,
  isbn = {9798400703850},
  address = {New York, NY, USA},
  pages = {1183–1199},
  numpages = {17},
  keywords = {temporal graph neural networks, dynamic graphs, programming framework, data abstractions},
  location = {La Jolla, CA, USA},
  series = {ASPLOS '24}
}

ICML

Ithemal: Accurate, Portable and Fast Basic Block Throughput Estimation using Deep Neural Networks

Charith Mendis, Alex Renda , Saman P. Amarasinghe , and Michael Carbin

In Proceedings of the 36th International Conference on Machine Learning (ICML) , Mar 2019

Best Paper award at the ML for systems workshop co-located with ISCA’19

PDF Bib Code

@inproceedings{ithemal-icml2019,
  author = {Mendis, Charith and Renda, Alex and Amarasinghe, Saman P. and Carbin, Michael},
  editor = {Chaudhuri, Kamalika and Salakhutdinov, Ruslan},
  title = {Ithemal: Accurate, Portable and Fast Basic Block Throughput Estimation
                 using Deep Neural Networks},
  booktitle = {Proceedings of the 36th International Conference on Machine Learning (ICML)},
  series = {Proceedings of Machine Learning Research},
  volume = {97},
  pages = {4505--4515},
  publisher = {{PMLR}},
  year = {2019},
  url = {http://proceedings.mlr.press/v97/mendis19a.html},
  timestamp = {Tue, 11 Jun 2019 15:37:38 +0200},
  biburl = {https://dblp.org/rec/conf/icml/MendisRAC19.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org},
  award = {Best Paper award at the ML for systems workshop co-located with ISCA'19},
  bib = {https://dblp.org/rec/conf/icml/MendisRAC19.html?view=bibtex},
}

Funding

I am truly thankful for the sponsors of our research at the ADAPT lab. In particular, we are partially funded by the ACE center, one of the seven centers in JUMP 2.0, a Semiconductor Research Corporation (SRC) program sponsored by DARPA, by NSF (including CAREER Award), by DARPA (including Young Faculty Award), by IIDAI and through generous gifts and cloud computing resources from Google, Intel, Amazon and Qualcomm.