Rishit Dagli

I also write a blog.

I am a CS Sophomore at The University of Toronto. I love researching and working with Machine Learning, especially Computer Vision. Coming from the software and robotics background, I contribute extensively to/ maintain popular open-source projects like TensorFlow, PyTorch Foundation, Kubernetes, Kubeflow, PapersWithCode, freeCodeCamp among others. I also love building open-source projects (usually related to Kubernetes and Machine Learning), some of which have been pretty popular which could be found on my GitHub. Seeing my work at a rather young age, I was invited to speak at 2 TEDx and 1 TED-Ed events. In a previous life I used to do a lot of robotics and software development. Furthermore, I have also represented my country in international olympiads. Feel free to talk with me about anything CS, Math, Robotics, or Physics.

I am advised by Prof. David Lindell as a part of the DGP Lab, Vector Institute at the University of Toronto where I research Diffusion. I am currently working as a Research Intern at Civo Cloud researching vision and multimodal models where I am advised by Josh Mesout.

My research community service can be summarized as being a: program committee member for ICLR Tiny Papers 2023; reviewer for ICLR PML4DC 2023, NeurIPS 2023, Cloud Native Wasm Day, and ICLR 2024.

As to why I got drawn to Computer Vision, I do believe it is one of the richest modalities but also read this excerpt when I was quite young, from the book, “Visual Reconstruction” by Andrew Blake and Andrew Zisserman, some of my favourite researchers:

We count it a great privilege to be working in a field as exciting as Vision. On the one hand there is all the satisfaction of making things that work - of specifying, in mathematical terms, processes that handle visual information and then using computers to bring that mathematics to life. On the other hand there is a sense of awe (when time permits) at the sheer intricacy of creation. Of course it is the Biological scientists who are right in there; but computational studies, in seeking to define Visual processes in mathematical language, have made it clear just how intrinsically complex must be the chain of events that constitutes “seeing something”.

I deeply thank the following organization for current/ in the past supporting my work either through scholarships or research grants or support of some sort: Linux Foundation, Google AI, Google Cloud, University of Toronto, Vector Institute, Stanford, CNCF, and Intel AI.

news

Feb 13, 2024	Received the T-CAIREM award for students at UofT.
Aug 17, 2023	1 oral + 1 poster accepted to PyTorch Conference.
Mar 15, 2023	Soon joining Civo, a startup as one of the first ML research scientists.
Mar 2, 2023	1 paper accepted to ICLRW.
Jan 17, 2023	Recipient of the Google AI Research Grant for 2023.

selected publications

Tuning In

: Analysis of Audio Classifier Performance in Clinical Settings with Limited Data

Hamza Mahdi, Eptehal Nashnoush, Rami Saab, and 4 more authors

Feb 2024

Bib HTML PDF

@misc{mahdi2024tuning,
  title = {Tuning In <img src="assets/img/emojis/audio.png" alt="custom_emoji" style="width: 20px; height: 20px;"> : Analysis of Audio Classifier Performance in Clinical Settings with Limited Data},
  author = {Mahdi, Hamza and Nashnoush, Eptehal and Saab, Rami and Balachandar, Arjun and Dagli, Rishit and Perri, Lucas X. and Khosravani, Houman},
  year = {2024},
  month = feb,
  eprint = {2402.10100},
  archiveprefix = {arXiv},
  primaryclass = {cs.SD},
}

DiffuseRAW: End-to-End Generative RAW Image Processing for Low-Light Images

Rishit Dagli

Feb 2023

Bib HTML PDF

@misc{dagli2024diffuse,
  title = {<span style="color: #80ED12;">D</span><span style="color: #92E10B;">i</span><span style="color: #A5D604;">f</span><span style="color: #B6C602;">f</span><span style="color: #C7B601;">u</span><span style="color: #D5A405;">s</span><span style="color: #E39209;">e</span><span style="color: #EC7F12;">R</span><span style="color: #F66C1C;">A</span><span style="color: #EC7F12;">W</span>: End-to-End Generative RAW Image Processing for Low-Light Images},
  author = {Dagli, Rishit},
  year = {2023},
  eprint = {2402.18575},
  archiveprefix = {arXiv},
  primaryclass = {eess.IV},
}

PyTorch Made Efficient for the Edge: WASI-NN

Rishit Dagli, and Shivay Lamba

In PyTorch Conference, Oct 2023

Bib HTML Poster

@inproceedings{dagli2023ptc,
  author = {Dagli, Rishit and Lamba, Shivay},
  title = {PyTorch Made Efficient for the Edge: WASI-NN},
  booktitle = {PyTorch Conference},
  year = {2023},
  month = oct,
}

Orchestrating Machine Learning on Edge Devices with PyTorch and WebAssembly (Oral)

Rishit Dagli, and Shivay Lamba

In PyTorch Conference, Oct 2023

Bib HTML

@inproceedings{dagli2023ptcoral,
  author = {Dagli, Rishit and Lamba, Shivay},
  title = {Orchestrating Machine Learning on Edge Devices with PyTorch and WebAssembly (Oral)},
  booktitle = {PyTorch Conference},
  year = {2023},
  month = oct,
}

Astroformer: More Data Might Not be All You Need for Classification

Rishit Dagli

In International Conference on Learning Representations Workshops, Apr 2023

Bib HTML PDF

@inproceedings{dagli2023astroformer,
  author = {Dagli, Rishit},
  title = {Astroformer: More Data Might Not be All You Need for Classification},
  booktitle = {International Conference on Learning Representations Workshops},
  year = {2023},
  month = apr,
}

CPPE-5: Medical Personal Protective Equipment Dataset

Rishit Dagli, and Ali Mustufa Shaikh

SN Computer Science, Mar 2023

Bib HTML PDF

  author = {Dagli, Rishit and Shaikh, Ali Mustufa},
  title = {CPPE-5: Medical Personal Protective Equipment Dataset},
  journal = {SN Computer Science},
  year = {2023},
  month = mar,
  day = {16},
  volume = {4},
  number = {3},
  pages = {263},
  issn = {2661-8907},
  url = {https://doi.org/10.1007/s42979-023-01748-7},
}

Deploying a smart queuing system on edge with Intel OpenVINO toolkit

Rishit Dagli, and Süleyman Eken

Soft Computing, Aug 2021

Bib HTML PDF

@article{Dagli2021,
  author = {Dagli, Rishit and Eken, S{\"u}leyman},
  title = {Deploying a smart queuing system on edge with Intel OpenVINO toolkit},
  journal = {Soft Computing},
  year = {2021},
  month = aug,
  day = {01},
  volume = {25},
  number = {15},
  pages = {10103-10115},
  issn = {1433-7479},
  url = {https://doi.org/10.1007/s00500-021-05891-2},
}