1. Programming Portfolio
Building a Smart Dashboard that Utilizes OpenAI's Large Language
Model (LLM) to Create a ChatBot that Integrates with Oil and Gas
Data (2023)
• Written in Python leveraging Dash, Plotly, Geopandas, Leaflet, FastAPI,
OpenAI, and LangChain
• Created database and integrated it using PostgreSQL with map visualization
on real-time,
• Built interactive map using Leaflet with features of well log pop-up, ESRI
layout, and advanced filter,
• Leveraging OpenAI's Large Language Model to build an intelligent assistant
that can ask everything about the data, create plots, and create summaries
without any querying skills
2. Programming Portfolio
Big Web Application for PT Pertamina Hulu Mahakam using Python
and Streamlit for Well Log and Seismic Data (2022)
• A big web application for PT Pertamina Hulu Mahakam project with several
features including coordinate and TVDSS interpolation, lumping editor,
database crude system using NoSQL database (MongoDB), Exploratory
Data Analysis (single and multiple well logs), and well log prediction,
• Written on Python with utilizing Streamlit, Numpy, Pandas, Seaborn, Plotly,
Sckit-learn, Missingno, PyMongo, PyProj, XGBoost, Optuna, and Joblib
library,
• Created a login system for geoscience engineers of PT Pertamina Hulu
Mahakam using only Streamlit and MongoDB server,
• Handling 300+ well logs using Pandas, Interpolating the well trajectory for
incomplete well logs using Numpy, and transforming coordinate into latitude
3. Programming Portfolio
and longitude using PyProj,
• Created a database for well log and seismic data using NoSQL database
(MongoDB),
• Well log parameter prediction (PHIE, NPHI, RHOB) using XGBoost and
Optuna hyperparameter optimization. Achieving 0.016 of MAE score.
DT Log Prediction using Gradient Boosting Algorithm in Jatibarang
Field (2022)
• Python machine learning project to predict DT log parameters based on other
parameters from well logs in Jatibarang field,
• Written on Python with utilizing Numpy, Pandas, Lasio, Welly, Seaborn,
Plotly, Matplotlib, Sckit-Learn, and Missingno library,
• 14 training wells and 1 blind test well with the parameters for the training
data are CALI, GR, ILD, NPHI, RHOB, SP, and DT,
• the pre-process includes transforming, using logarithmic and Yeo-Johnson
transformations and removing outlier data using one-class SVM,
• using GridSearchCV for the hyperparameter optimization stage,
• Produces an error value of 10.46% based on MAPE calculation with an R2
score of 0.72 and processing time is only 9.36 seconds.
4. Programming Portfolio
FORCE 2020 Well Log Challenge Dashboard (2022)
• Well log dashboard using open-source data from FORCE 2020 Well Log
Challenge. It consists of well log plots, 2D and 3D distributions, well
positions based on lithology and log parameters, data table, box-plots,
rug plots and interactive dropdowns for each plot,
• Written on Python with utilizing Numpy, Pandas, Plotly, Dash, Dash
Bootstrap Component (DBC), Gunicorn,
• Deployed on Heroku-App so that everyone can easily access the
dashboard, accessible via the link https://force2020-dash.herokuapp.com.
5. Programming Portfolio
My explanation and code guide can be seen in my writing on
Medium https://medium.com/@naharirasif and GitHub
https://github.com/nrasif
Also, I have an Instagram page to write what I have learned
from the internet https://www.instagram.com/insightiq.id/