Trying to Understand PaperTrade's Tokenomics
How PaperTrade pays winners, rewards losses with PAPER, and gets people to participate.
Writing
Blog posts on product work and business systems, alongside technical notes on mathematics, AI, and research. Browse the two collections below.
Blog posts
Product decisions, business systems, and lessons from building things.
How I Define When a Payment Is Complete
The settlement boundary, its trade-off, and the acceptance criteria I work through with engineers.
TRON Energy Is a Product Requirement
Why TRON resource management changes the payment flow and the user experience.
Why We Built an Internal CRM After Trying Salesforce
How I map sales, delivery, and finance handovers for an internal CRM at Quandatics.
A Web3 Payment Needs More Than a Transaction Hash
The product decisions behind payment states, confirmation, and settlement in a Web3 gateway.
Talentbank Boardroom Challenge 2025
A small photo record from the Talentbank Boardroom Challenge 2025.
A Small Talk About Hackathons
What repeated hackathons taught me about exhaustion, comparison, and taking a proper break.
Dead Internet Theory #1
Why polished posts can feel disconnected from the people behind them.
Building a DeepSeek-R1 Chatbot
A Gradio chatbot experiment with visible intermediate output and the 1.5B distilled DeepSeek-R1 model.
Eco Finance: A PayHack Prototype
A carbon-footprint and financial-literacy prototype built during PayHack 2024 with OpenFinance APIs.
First Physical Hackathon Experience
What I learned from my first physical hackathon at PayHack 2024, where we used OpenFinance APIs to prototype a financial-literacy app.
Technical notes
Mathematical explanations, derivations, and worked examples in AI and research.
Linear regression8 lessons · from fitting a line to statistical testing
Follow the derivations in order. Each lesson builds on the notation and results from the previous one.
- Simple Linear RegressionAn introduction to simple linear regression, the OLS method, and deriving the estimators using Sxx and Sxy notation.5 min
- Multiple Linear RegressionExtending simple linear regression to multiple predictors using matrix notation and deriving the OLS estimator in matrix form.5 min
- Polynomial RegressionUnderstanding polynomial regression as a special case of multiple linear regression, with design matrix construction and overfitting considerations.5 min
- CorrelationDefining the Pearson correlation coefficient using Sxx, Syy, Sxy, Sx, and Sy notation, and exploring its properties.4 min
- R-squared: SST, SSE, SSR and the Relationship with CorrelationDefining the coefficient of determination from the sum of squares decomposition and proving that R-squared equals the square of the correlation coefficient.7 min
- Standard Error in RegressionDeriving the standard errors of the regression coefficients using MSE, Sxx, and the variance-covariance matrix.5 min
- Confidence Intervals for Regression CoefficientsConstructing confidence intervals for beta_0, beta_1, and beta_i using the t-distribution and standard errors.5 min
- Statistical Testing in RegressionPerforming hypothesis tests on regression coefficients using t-tests, p-values, and the connection to confidence intervals.5 min
CryptoRails: From Deposit to Settlement
Deposit detection, consolidation, and verified-webhook settlement across EVM and TRON.
Why Can't Time Flow Backwards? Entropy, Yu Deng, and What the Mathematics Actually Proves
A first-principles look at why reversible particle collisions can produce an irreversible Boltzmann equation, how entropy enters mathematically, and what Yu Deng's Fields Medal work with Zaher Hani and Xiao Ma actually proved.
The Fokker-Planck Equation Explained, from an SDE to the Density it Evolves
A first-principles walkthrough from a single stochastic trajectory to the Fokker-Planck equation for its probability density, derived from Ito's lemma, with a from-scratch Python solution and an Ornstein-Uhlenbeck example checked three ways.
An Introduction to MLOps: A Complete Tutorial with a Hands-On Example
A plain-language tutorial that follows the Intro to MLOps slides end to end: reproducibility, the software cycle, DevOps, the ML lifecycle, MLOps, and MLflow, with diagrams and a runnable hands-on.
Tree-Based Models Explained: From Decision Trees to Random Forest and XGBoost
An intuition-first but math-heavy walkthrough from decision tree splits to bagging in Random Forests and boosting in XGBoost.
Fast Fourier Transform Explained from the DFT, with a Sound-Wave Example
A first-principles walkthrough from sampled sound waves to the DFT, then from the DFT to the Cooley-Tukey FFT, with a from-scratch Python implementation and complexity proof.
Softmax Regression Explained from Logistic Regression
A first-principles walkthrough from odds and log-odds to logistic regression, then from logistic regression to softmax regression.
Statistical Testing in Regression
Performing hypothesis tests on regression coefficients using t-tests, p-values, and the connection to confidence intervals.
Confidence Intervals for Regression Coefficients
Constructing confidence intervals for beta_0, beta_1, and beta_i using the t-distribution and standard errors.
Standard Error in Regression
Deriving the standard errors of the regression coefficients using MSE, Sxx, and the variance-covariance matrix.
R-squared: SST, SSE, SSR and the Relationship with Correlation
Defining the coefficient of determination from the sum of squares decomposition and proving that R-squared equals the square of the correlation coefficient.
Correlation
Defining the Pearson correlation coefficient using Sxx, Syy, Sxy, Sx, and Sy notation, and exploring its properties.
Polynomial Regression
Understanding polynomial regression as a special case of multiple linear regression, with design matrix construction and overfitting considerations.
Multiple Linear Regression
Extending simple linear regression to multiple predictors using matrix notation and deriving the OLS estimator in matrix form.
Simple Linear Regression
An introduction to simple linear regression, the OLS method, and deriving the estimators using Sxx and Sxy notation.
Solar Photovoltaic Systems for Wastewater Treatment Plants in Malaysia
A review of energy demand, solar integration, forecasting, and control considerations for Malaysian wastewater treatment plants.
Gradient Descent Algorithm Explained
The gradient descent update rule, worked examples, and the effect of the learning rate.
Three Questions for Reinforcement Learning in Healthcare
A practical reading of three safety questions for reinforcement learning in clinical settings: confounding, effective sample size, and prospective behaviour.
Action-value methods with incremental step size in reinforcement learning
Deriving the incremental update rule for action-value estimation in k-armed bandits, and why constant step sizes suit nonstationary problems.
No writing matches this selection.