Back to Portfolio

Tutorials & Deep Dives

Comprehensive guides, from-scratch implementations, curated paper collections, and hands-on tutorials across LLMs, diffusion models, computer vision, and ML systems.

40+
Resources
8
Categories
1.2K+
GitHub Stars

LLMs & Fine-Tuning

From fundamentals to scaling laws — everything about large language models

LLM Fine-Tuning: A Complete Guide

Guide

A comprehensive, chapter-by-chapter guide to LLMs — from probability basics to scaling laws, with hands-on fine-tuning code and exercises.

LLM Fine-Tune Desktop App

Tutorial

PyQt5 desktop GUI for fine-tuning, evaluating, and deploying LLMs using torchtune — no command-line required. Full visual workflow.

LLM Architectures & Deployment

Guide

Comprehensive guides for working with Large Language Models — architectures, training pipelines, and deployment strategies explained in depth.

LoRA: From Scratch Implementation

Implementation

Low-Rank Adaptation of Large Language Models — parameter-efficient fine-tuning implemented from scratch with detailed code walkthrough.

Efficient Inference & Compression

KV cache optimization, model quantization, and LLM compression research

KV Cache & LLM Compression Papers

150+ Papers

Curated collection of 150+ research papers on KV Cache Management, KV Cache Compression, and LLM Compression for efficient inference.

Awesome Model Quantization

Papers

Curated list of papers, docs, and code about model quantization — aimed at providing comprehensive info for quantization research.

YOLOX Quantization & Distillation

Tutorial

Hands-on tutorial combining YOLOX with Quantization Aware Training and Knowledge Distillation for efficient real-time object detection.

ONNX: Complete Tutorial

Guide

A comprehensive, book-style tutorial covering everything about ONNX — from fundamentals to production deployment and optimization.

Diffusion Models

Image generation, text diffusion, and denoising from theory to implementation

Diffusion Language Models

Implementation

Text generation using diffusion-style denoising — iteratively refining noisy sequences into coherent text, an alternative to autoregressive LLMs.

Reparameterization Denoising

Implementation

Implementation of diffusion-based denoising with reparameterization — organized and shared with detailed code walkthrough.

Diffusion Models: Survey & Taxonomy

Papers

Comprehensive survey and taxonomy of diffusion model papers — organized by architecture, application, and training methodology.

Attention & Transformers

Efficient attention mechanisms, linear attention, and transformer architecture surveys

Attention Mechanisms: 3 Surveys

Guide

Three in-depth surveys covering efficient transformer architectures, attention variants, and optimization techniques for scalable inference.

REGLA: Gated Linear Attention

Implementation

Refining Gated Linear Attention — efficient alternative to softmax attention for scalable sequence modeling with linear complexity.

From PCA to VAE

Tutorial

Understanding dimensionality reduction from classical PCA through autoencoders to Variational Autoencoders — theory and implementation.

Computer Vision

Image enhancement, object detection, segmentation, and low-level vision

Low-Level Vision: Complete Guide

Guide

Super-Resolution, denoising, deblurring, dehazing, low-light enhancement, artifact removal — end-to-end models with PSNR/SSIM benchmarks.

FlashDet: End-to-End Detection

Tutorial

Complete training system with PyQt5 desktop app — LoRA/QLoRA fine-tuning, knowledge distillation, ONNX export, INT8 quantization. 100+ FPS.

YOLOv8 PyTorch Implementation

Implementation

Modular PyTorch implementation of YOLOv8 for object detection with clear architecture, training pipelines, and detailed documentation.

Feature Detection from Scratch

Implementation

From-scratch implementations of classic feature detection and description algorithms — SIFT, SURF, ORB, Harris, and more.

Image Object Removal & Inpainting

Tutorial

Remove objects from photos including shadows and reflections using generative inpainting — end-to-end diffusion-based restoration.

Bayer Low-Light Enhancement

Implementation

Low-light image enhancement directly on Bayer pattern data — RAW image processing with deep learning for mobile camera pipelines.

ML Systems & Production

System design, data drift, monitoring, and production ML architecture

ML System Design Guide

Guide

Comprehensive guide to ML System Design — LLM serving, training pipelines, scaling, and real-world architecture patterns for interviews and practice.

Data Drift in Production ML

200+ Papers

Research on data drift — taxonomy (covariate/concept/label shift), mathematical formulations (KL, PSI, Wasserstein), monitoring architectures, and 200+ curated papers.

Foundations & Mathematics

ML research foundations, linear algebra, DSA, and courses

ML Researcher Foundations

Roadmap

Structured, end-to-end roadmap for becoming a strong ML researcher — mathematical foundations, ML theory, deep learning, optimization, and efficient ML.

Data Structures & Algorithms

170+ Stars

Well-organized repository covering core DSA from fundamentals to advanced topics — systematic learning, coding interviews, and reference implementations.

CV Interview Questions

Guide

Curated computer vision interview questions covering CNNs, object detection, segmentation, GANs, transformers, and production deployment.

Curated Reading Lists

Awesome lists, paper collections, and survey compilations

Low-Level Vision Paper Record

Papers

Curated record of papers on low-level vision tasks — super-resolution, denoising, deblurring, and image restoration research.

Awesome Transformer Attention

Papers

Curated list of research papers on efficient attention mechanisms and transformer architectures for NLP and vision.

Awesome VLM Architectures

Papers

Curated collection of Vision-Language Model architectures — from CLIP to GPT-4V, covering multi-modal learning and alignment.

OCR Research Papers

Papers

Collection of OCR research papers — scene text detection, recognition, and end-to-end document understanding systems.