Hi, Ritesh here.
I’m a machine learning engineer focused on building and understanding practical deep learning systems. I previously worked at Zomato and Samsung R&D Institute India.
I use this site to document what I learn while building machine learning projects.
Projects
nanoFM
A minimal PyTorch implementation of flow matching, connecting the mathematical formulation to training and sampling.
nanoSBGM
A minimal PyTorch implementation of score-based generative modeling, exploring SDE formulations and stochastic and deterministic sampling.
Qwen2.5 SFT
Fine-tuning Qwen2.5-0.5B for tool calling with the xLAM dataset, with LoRA training and structured evaluation.
Achieved 93.89% tool accuracy and 98.75% valid JSON on 400 held-out test samples.
nanoOCR
A minimal OCR pipeline with DBNet text detection, CRNN text recognition, and ONNX inference with INT8 quantization.
The recognizer achieved 68.9% exact word-match accuracy on the IIIT5K test set.
Experience
Machine Learning Engineer II
Zomato, Gurugram
June 2026 – Aug 2026
Machine Learning Engineer
Samsung R&D Institute India, Noida
Jan 2024 – June 2026
Machine Learning Engineer Intern
Samsung R&D Institute India, Noida
Jan 2023 – Dec 2023