Muhammad Daffa Robani

Experience

• 

  Institut Teknologi Bandung

  Feb 2017 - Mar 2021

  Conducted scientific researches on the utilization of statistical concepts in engineering problems and prepared academic paper for international conferences. I contributed in developing in-house laboratory's MATLAB code on surrogate modeling framework, prepared two international academic papers, and assisted students in conducting their research. Assisted several lecturers in providing additional lessons to enhance the students' understanding of the courses. I contributed in creating experiment module by using MATLAB GUI to substitute physical experiment procedure, creating statistic course module on Octave and Python codes, and managing also conducting weekly tutorials and experiments for various engineering courses.

  • Research Assistant

    Aug 2019 - Mar 2021

  • Teaching Assistant

    Feb 2017 - Jun 2020
• 

  Gexcon

  Jun 2018 - Jul 2018

  Intern

  Project: Determine the potential hazard of a petrochemical plant in Indonesia by using GEXCON's computational fluid dynamics (CFD) software.

• 

  DAX Indonesia

  Apr 2021 - Nov 2021

  Data Scientist

  Managing and maintaining an end-to-end data processing framework to provide data-driven analysis and recommendation in optimizing marketing efforts of business clients. Daily tasks:- Configuring tags in the client's product website by using Google Tag Manager and Javascript.- Analyzing digital activities data by using SQL via Google BigQuery, Google Analytics, and Python programming language.- Create ML models to learn customer behaviours and propose marketing recommendations.- Present analysis regarding marketing performance to the clients and creating marketing report dashboard by using Google Data Studio.Tools: Python, MySQL via Google BigQuery, Google Analytics, Google Tag Manager, JavaScript and Google Data Studio.Projects: Explainable lead scoring model for an automotive industry client, cluster segmentation analysis for automotive and food nutrition clients, and Markov-Chain marketing attribution model. Show less

• 

  AwanTunai

  Dec 2021 - now

  - Lead a team of five data scientists in developing and deploying machine learning models for risk process automation and improvement. - Establish and maintain feature library as an extendable Python package, compiling features derived from transaction, bank statement, and loan data for easy use in downstream tasks.- Develop and maintain ETL tasks in Airflow for features computation and batch prediction. - Key individual projects:RNN with attention mechanism for basket recommendation system: Built an RNN-based model to predict a set of marketable items for a merchant's next inventory purchase, where attention mechanism is used to help the model put different importance of the item attributes. This system increased the merchant's revenues by 16%. Automated knowledge graph insights extraction: Constructed a knowledge graph from historical semi-structured underwriting data, and developed a module to automatically extract interesting insights based on the node attributes and graph structure. This resulted in an average 3.5/5 rating from data analysts for the extracted insights. Interpretable credit approval model: Developed a credit approval model based on financial report, bank statement, and sales data. The model's behaviour was translated into heuristic rules, creating an easily interpretable framework. The model achieved 40% automation rate with 0% false approval rate, the model is now in production. - Key team projects: Invoice and bank statement OCR, supplier risk monitoring tool based on bank statement and sales data, fully automated small size line of credit approval model, and people counting from CCTV records. - Other individual projects: Transformer model for financial data summarization, fraudulent transaction detection model, and transfer back detection model. Show less

  • Senior Data Scientist

    Nov 2024 - now

  • Data Scientist

    Jul 2022 - Nov 2024

  • Associate Data Scientist

    Dec 2021 - Jun 2022