ML & AI Solutions Architect | Scalable ETL and API Deployment | Survey & Digital Strategy Innovator | Army Reserved Officer 🎖

Meet Fahmi Zainal, a visionary Data Scientist with a passion for turning raw data into actionable insights and strategies that drive business growth. With expertise spanning machine learning, data engineering, and ETL pipelines, Fahmi has successfully delivered impactful solutions in domains like digital marketing, survey optimization, and business intelligence.

Currently, Fahmi leads innovative projects at INVOKE Solutions, specializing in ROAS optimization, geocoding pipelines, and automated survey systems. As a lifelong learner and seasoned mentor, he thrives on empowering teams and shaping the future of AI-driven insights.

• 🏅 First Place for developing a fine-tuned LLM Judge under computational constraints.
• Collaborated with a cross-functional team to deliver a solution praised for its practicality and creativity.

• Ranked Top 28% out of 1,908 teams globally.
• Developed an ensemble model with 87% ROC-AUC accuracy leveraging LightGBM, XGBoost, and Optuna for hyperparameter tuning.

• Built a deep learning model with 84% accuracy for flood forecasting using weather datasets.
• Deployed the solution for real-time predictions supporting disaster management initiatives.

• Developed a web-based platform integrating Streamlit, Shiny Apps, and FastAPI for survey processing.
• Reduced operational time by 95% through streamlined data cleaning, weighting, and visualization workflows.

• Designed predictive models and dashboards for digital campaign optimization, increasing revenue by 50%.
• Engineered ETL pipelines with Databricks and MongoDB for seamless data processing.

• Movie Recommender System: Built a content-based filtering model to recommend movies using metadata features such as genre, cast, and director.
• Job Recommender System: Developed a hybrid model combining collaborative filtering and content-based filtering to suggest tailored job opportunities.
• Ads Recommender: Engineered a machine learning model for ad campaign optimization, leveraging historical user behavior to rank ad sets by effectiveness.
• All systems deployed as Streamlit web apps for interactive user experiences, with backend integrations using Python and FastAPI.

• Developed an API using YOLOv5 for automated detection of Metisa plana pupae in images.
• Integrated with Streamlit for user interaction and deployed using Docker.

For the full list of projects, visit My Portfolio.

For the full list, visit the Certifications section on My Portfolio.

WakaTime Stats

GitHub Stats