CV
Contact Information
| Name | Utsav Akhaury |
| utsav.akhaury.2015@gmail.com |
Experience
-
Mar 2026 – Present AI Software Engineer
Spot AI
San Francisco, CA
- Building video AI agents for real-time perception, reasoning, and action.
-
Sep 2021 – Feb 2026 Doctoral Assistant
Swiss Federal Institute of Technology, Lausanne (EPFL)
Lausanne, Switzerland
- Built deep learning systems for large-scale image restoration using data from NASA and ESA missions, delivering ~1000× faster inference than classical methods.
- Designed end-to-end ML pipelines from data simulation through model training, evaluation, and deployment — scaling large vision models across multi-node GPU clusters via Slurm-scheduled HPC workflows.
- Developed a novel multi-resolution image fusion framework that combines multi-telescope, multi-spectral imaging data.
- Mentored junior researchers on independent projects and collaborated with teams across Europe and North America.
-
May 2021 – Aug 2021 Open-source Developer
Google Summer of Code (GSoC)
Remote
- Built an automated framework for training ML surrogates that deliver near real-time predictions in place of computationally-expensive radar simulations — a GSoC contribution to gprMax (open-source simulation platform).
-
May 2021 – Aug 2021 Research Intern
Observatoire de la Côte d'Azur
Nice, France
- Built 3D CNNs extending a 2D point detection algorithm to 3D on a 1 TB signal-cube dataset.
- Designed a custom loss function with tunable intensity weighting to recover faint signals and improve detection of weak targets.
-
Nov 2020 – Apr 2021 Research Intern
French Alternative Energies and Atomic Energy Commission (CEA)
Paris, France
- Co-developed deep learning deconvolution algorithms for optical and radio imaging, marking the first application of this technology to radio-interferometry.
- Contributed to a synthetic dataset of 51,000+ images using Transformer-based generative models (Tensor2Tensor), released publicly as a community benchmark.
Education
-
Sep 2021 – Feb 2026 PhD in Physics
Swiss Federal Institute of Technology, Lausanne (EPFL)
Lausanne, Switzerland
- Thesis: Advancing astronomical image deconvolution with deep learning
- Supervisors: Pascale Jablonka, Jean-Luc Starck
-
Aug 2016 – May 2021 Dual Degree: M.Sc. Physics + B.E. Electronics & Instrumentation Engineering
Birla Institute of Technology and Science (BITS), Pilani
Pilani, India
- Thesis: Reconstruction of multi-frequency radio galaxy images using Machine Learning & ADMM.
- Supervisor: Jean-Luc Starck
Skills
Languages Python C++ Julia MATLAB Rust
ML / Deep Learning PyTorch CUDA Hugging Face TensorFlow scikit-learn
Computer Vision OpenCV ONNX TensorRT FFmpeg
MLOps & Tooling Slurm Docker Kubernetes Google Cloud GitHub Actions
Awards
-
2023 EPFL PhD Excellence Fellowship
Swiss Federal Institute of Technology, Lausanne (EPFL)
Selected as one of 27 PhD students across all departments.
-
2023 Young Scientist Support Grant
Swiss Society for Astrophysics and Astronomy (SSAA)
Travel grant awarded to support academic visits, conferences, and summer schools.
-
2017 BITS Merit-cum-Need Scholarship
Birla Institute of Technology and Science (BITS), Pilani
Awarded to top-performing students at BITS Pilani. Covered 80% of annual tuition.
Service and Volunteering
-
2026 Journal Reviewer
Astronomy & Astrophysics (A&A)
Peer-reviewed manuscripts at the intersection of deep learning and astronomy.
-
2025 STEM Outreach Volunteer
International School of Lausanne
Shared my work on AI and space telescopes with school students on the International Day of Women and Girls in Science.
-
2023–2024 External Relations Committee Member
EPFL Data Analytics Group
Managed corporate-partner outreach and sponsorship for EPFL’s data analytics community.
-
2010–2024 Astronomy Outreach Organizer
Observatoire de Genève BITS Pilani The Scindia School
Organized astronomy outreach events and public sky-watching sessions, engaging audiences ranging from school students to working professionals.