This study investigates the adsorption behavior of Congo red (CR) onto activated biochar (ABHC) derived from Haematoxylum campechianum waste. The ABHC was prepared by soaking the precursor in phosphoric acid and then pyrolyzing it. The material was characterized using SEM/EDS, BET, XRD, and FTIR, and its pHpzc was determined. Batch experiments were conducted under varying pH (4–10), temperature (300.15–330.15 K), and adsorbent dose (1–10 g/L). The adsorption isotherms were evaluated to determine the maximum adsorption capacity (Qmax), and kinetic studies were performed at two initial concentrations (25 and 50 mg/L) and a maximum contact time of 48 h. The reusability of ABHC was assessed through adsorption-desorption cycles. The Langmuir model predicted a maximum adsorption capacity of 114.8 mg/g at 300.15 K, pH 5.4, and a dose of 1.0 g/L. The study also highlights the application of advanced machine learning techniques to optimize the adsorption process. A Gradient Boosting regression model was developed and fine-tuned using Bayesian optimization. The model efficiently navigated the input space to maximize the removal percentage, achieving a predicted efficiency of approximately 90.47% under optimal conditions. The results show that ABHC has a high adsorption capacity for CR, with Qmax values ranging from 10.51 to 114.8 mg/g depending on pH. The adsorption capacity decreased at higher pH values due to repulsion between the negatively charged surface of ABHC and the negatively charged CR molecules. The adsorption isotherms were best fitted by the Langmuir and Redlich–Peterson models, with the Redlich–Peterson model showing the best fit across the studied pH range. The adsorption capacity decreased with increasing ABHC dose due to the reduction in active sites. The adsorption capacity was also influenced by temperature, with lower temperatures favoring adsorption. The study also evaluated the reusability of ABHC, showing that it could be desorbed using NaOH, although the desorption efficiency was low. The machine learning model demonstrated high predictive accuracy, with R² values of 0.99 for the training set and 0.914 for the testing set. The model's performance metrics indicated that it could accurately predict the removal percentage, although there was a slight underprediction bias. The study concludes that ABHC is a promising adsorbent for CR removal, and the integration of machine learning techniques can significantly enhance the efficiency of adsorption processes.This study investigates the adsorption behavior of Congo red (CR) onto activated biochar (ABHC) derived from Haematoxylum campechianum waste. The ABHC was prepared by soaking the precursor in phosphoric acid and then pyrolyzing it. The material was characterized using SEM/EDS, BET, XRD, and FTIR, and its pHpzc was determined. Batch experiments were conducted under varying pH (4–10), temperature (300.15–330.15 K), and adsorbent dose (1–10 g/L). The adsorption isotherms were evaluated to determine the maximum adsorption capacity (Qmax), and kinetic studies were performed at two initial concentrations (25 and 50 mg/L) and a maximum contact time of 48 h. The reusability of ABHC was assessed through adsorption-desorption cycles. The Langmuir model predicted a maximum adsorption capacity of 114.8 mg/g at 300.15 K, pH 5.4, and a dose of 1.0 g/L. The study also highlights the application of advanced machine learning techniques to optimize the adsorption process. A Gradient Boosting regression model was developed and fine-tuned using Bayesian optimization. The model efficiently navigated the input space to maximize the removal percentage, achieving a predicted efficiency of approximately 90.47% under optimal conditions. The results show that ABHC has a high adsorption capacity for CR, with Qmax values ranging from 10.51 to 114.8 mg/g depending on pH. The adsorption capacity decreased at higher pH values due to repulsion between the negatively charged surface of ABHC and the negatively charged CR molecules. The adsorption isotherms were best fitted by the Langmuir and Redlich–Peterson models, with the Redlich–Peterson model showing the best fit across the studied pH range. The adsorption capacity decreased with increasing ABHC dose due to the reduction in active sites. The adsorption capacity was also influenced by temperature, with lower temperatures favoring adsorption. The study also evaluated the reusability of ABHC, showing that it could be desorbed using NaOH, although the desorption efficiency was low. The machine learning model demonstrated high predictive accuracy, with R² values of 0.99 for the training set and 0.914 for the testing set. The model's performance metrics indicated that it could accurately predict the removal percentage, although there was a slight underprediction bias. The study concludes that ABHC is a promising adsorbent for CR removal, and the integration of machine learning techniques can significantly enhance the efficiency of adsorption processes.