🔭 I’m currently working on Weather Image restoration task at the CVPR Lab (SCSS) at Trinity College Dublin, Ireland - building GenAI pipelines (Diffusion Transformers, Dynamic Token Routing, Token reconstruction), achieving a PSNR of 31dB on restored images.
👥 I’m looking to collaborate on Open-source ML tooling, MLOps infrastructure, and applied Computer Vision / Generative AI projects that move models from prototype to production.
🤝 I’m looking for work on Scaling large generative models in production, dataset curation and annotation workflows for vision, and cost-efficient inference strategies.
🌱 Currently learning — hardware-aware machine learning, including PEFT/LoRA, model quantization and distillation; ML compilers and runtimes (LLVM, XLA, TVM, ONNX); CUDA-based parallel and multi-GPU training; and production-grade inference optimization using Triton, TensorRT, and Kubernetes.
⚡ I’m an undergrad at IIT Roorkee who implemented a Transformer from scratch and builds end-to-end ML systems — research + deployment.
🎯
Focusing
IIT Roorkee '27 | AI/ML & GenAI | contributer nanoVLM @huggingface | Built Stable Diffusion & LLaMA2 from scratch | Next.js/MongoDB dev
- Roorkee, Uttarakhand, India (247667)
-
15:13
(UTC +05:30) - https://shauryaomar.tiiny.site/
- in/shauryaiitr
- https://www.kaggle.com/surelytrue/code
- https://orcid.org/0009-0003-8850-1853
Highlights
- Pro
Pinned Loading
-
huggingface/nanoVLM
huggingface/nanoVLM PublicThe simplest, fastest repository for training/finetuning small-sized VLMs.
-
vehicle_insurance_proj
vehicle_insurance_proj PublicAWS‑powered end‑to‑end MLOps pipeline for vehicle insurance data ingestion, training & deployment
Jupyter Notebook
-
Stable_Diffusion
Stable_Diffusion PublicI have coded the Stable Diffusion Mofdel from scratch.
Jupyter Notebook
-
Flight_Delay_Analysis
Flight_Delay_Analysis PublicCritical Insights into Flight Delay Patterns
Jupyter Notebook
Something went wrong, please refresh the page to try again.
If the problem persists, check the GitHub status page or contact support.
If the problem persists, check the GitHub status page or contact support.

