A focused collection of articles spanning mechanistic interpretability, multimodal architectures, computer vision, transformer fine-tuning, and foundational ML algorithms.
How do modern AI frameworks train massive neural networks without collapsing under the weight of calculus? Explore primitives, input/weight Jacobians, vectorization, and reverse-mode AD.
Read More →Investigating whether Multimodal Sparse Autoencoders can identify, disentangle, and causally steer spatial reasoning features inside vision-language models without collateral semantic changes.
Read More →Extends YOLO+CSRNet crowd counting with a CNN+LSTM pipeline to forecast zone-level crowd density 5–10 minutes ahead and trigger proactive congestion alerts.
Read More →ML pipeline using RDKit, PyTorch, and XGBoost to predict Tg, FFV, Density, Tc, and Rg from SMILES strings. Ranked 656/2,400+ (65th on private leaderboard with XGBoost).
Read More →A POC that segments car parts with YOLOv12, classifies damage with a fine-tuned ResNet34, and generates a PDF damage report via Streamlit.
Read More →Fine-tuning large language models (LLMs) for code generation, such as Codex, StarCoder, and Code Llama.
Read More →An in-depth breakdown of recurrent cells, hidden state transitions, and step-by-step implementation of LSTM layers.
Read More →How the K-Nearest Neighbors algorithm makes predictions using vector space proximity and various distance metrics.
Read More →Exploring conditional probability, feature independence assumptions, and document classification using Naïve Bayes.
Read More →Understanding Shannon entropy, information gain, and tree pruning strategies for robust classification.
Read More →Diving into the sigmoid activation function, log-loss optimization, and gradient descent for binary classification.
Read More →A deep exploration into high-dimensional embedding spaces, vector representations, and semantic proximity.
Read on Medium →Implementing reinforcement learning basics, building a Q-learning policy network from scratch, and training autonomous decision agents.
Read on Medium →