Portrait of Arif Farhan Bukhori

Data Scientist at Singlife · Singapore

Arif Farhan
Bukhori.

I build machine learning and analytics systems that support business decisions, including customer propensity scoring in insurance and weekly price forecasting across 15 Indonesian provinces.

Experience

  • Singlife
  • PwC
  • JAPFA Group
  • Government of Lampung
  • Valores Global
  • ALPS Healthcare
  • Nanyang Business School

// about

About

I am a Data Scientist at Singlife in Singapore. My current work centres on a customer purchase-propensity model built on AWS SageMaker and Snowflake, one of the company's top three AI initiatives for 2026, together with the supporting data pipelines and dashboards.

Previously, I automated invoice processing and procurement reporting at JAPFA, developed a RAG-based policy assistant for the Office of the Governor of Lampung, and automated financial reporting at PwC. My focus is on taking models beyond the prototype stage and into systems that stakeholders rely on.

I hold an M.S. in Business Analytics from Nanyang Technological University, where I was named to the Dean's List as an ASEAN Development Scholar, and a B.S. in Computer Science (cum laude) from Universitas Gadjah Mada. I also lead analytics part-time for Pempek Tenda Biru, a multi-outlet restaurant chain in Indonesia.

Industry experience
3+ yrs
Business impact
$3M+
Work automated
6 FTE
Projects delivered
20+
Industries
7+
NTU CGPA, Dean's List
4.73/5

// experience

Where I've worked

Over three years of experience across insurance, agribusiness, government, healthcare and consulting.

  1. Mar 2026 – Present

    · Singapore

    Data Scientist

    Singlife

    Current

    Full-time since Jul 2026

    • Building a customer purchase-propensity model that scores 1M+ customers across 12 product categories, one of the company's top three AI initiatives for 2026.
    • Engineered 78+ features in Snowflake on a 4.8M-row dataset, trained XGBoost and LightGBM models on AWS SageMaker, and designed the A/B testing framework for a 1.5–2x conversion uplift target.
    • Contribute to lead routing for the advisory channel, compliance and sustainability data pipelines, and self-service Qlik dashboards.
    • Python
    • SQL
    • Snowflake
    • AWS SageMaker
    • XGBoost
    • Airflow
    • Qlik Sense
  2. Dec 2025 – Jan 2026

    · Singapore

    Data & Operations Analyst

    Valores Global

    Internship

    • Built a Python pipeline that generates audit reports for ALPS Healthcare across 170+ public healthcare institutions, reducing 17,200+ staff-minutes per quarter to near zero.
    • Designed and deployed four HR workflows in Power Automate and SharePoint covering task intake, employee records, leave management and payroll.
    • Python
    • Pandas
    • Power Automate
    • SharePoint
  3. Jun 2025 – Dec 2025

    · Indonesia

    Analyst to the Governor

    Lampung Provincial Government

    Contract role

    • Developed a RAG-based policy assistant over provincial regulations, reducing response preparation from several days to under 10 minutes per query.
    • Automated Tableau reporting for infrastructure and investment spending and standardised data from 20 agencies, reducing reporting workload by approximately 90%.
    • GenAI
    • RAG
    • Tableau
  4. Nov 2024 – Jun 2025

    · Indonesia

    Digital Transformation Analyst

    JAPFA Group

    Financial Shared Service division

    • Built 8 invoice ETL pipelines (Python and SharePoint to SAP), reducing processing time from 16 hours to 5 minutes and data entry errors by 95%.
    • Designed a procurement dashboard tracking invoice status, vendor performance and payment cycles across 300+ business units, shortening issue resolution by approximately 40%.
    • Conducted purchasing behaviour analysis and A/B tests on key product lines, identifying 4 bundle opportunities that informed pricing and vendor negotiations.
    • Python
    • SharePoint
    • SAP
    • ETL
    • A/B testing
  5. Mar 2023 – Present

    · Indonesia

    Lead Business Analyst

    Pempek Tenda Biru

    Part-time since Jun 2024

    • Drove 200% revenue growth in 7 months through campaigns based on sales analytics and customer segmentation, in partnership with 70+ influencers, 7 hotels and 100+ organisations.
    • Deployed an Odoo ERP across 5 outlets covering finance, HR and payroll, CRM, inventory and e-commerce, with role-based access for operations, finance and management.
    • Odoo
    • Python
    • PostgreSQL
    • Analytics
  6. Nov 2022 – Feb 2023

    · Indonesia

    Data Analyst

    PwC Indonesia

    Internship

    • Automated multi-entity financial reporting for a unicorn client, reducing the monthly close from 14 days to 3 hours.
    • Built a firm-wide ESG dashboard in Alteryx and Power BI with an NLP model that flags inconsistent disclosures, saving approximately 400 analyst hours per quarter.
    • Alteryx
    • Power BI
    • Scala
    • NLP

// projects

Selected work

Selected work in machine learning, GenAI and automation, alongside analytics and web projects. Select a project to view the details.

  • Featured · AI & ML Work project
    Customers scored
    1M+
    Product categories
    12
    Features engineered
    78+

    Singlife · Insurance propensity scoring across 12 product categories

    A model that predicts which existing customers are most likely to purchase an additional insurance product, so that campaigns prioritise the right customers. It is one of Singlife's top three AI initiatives on the 2026 company scorecard.

    View details

  • Featured · Data Science
    Annual savings
    $3M+
    MAPE
    3.2%
    Provinces
    15

    Chicken Price Forecasting System

    Production ML for the poultry industry across 15 Indonesian provinces

    Weekly chicken price forecasts for 15 provinces, used for production planning and generating $3M+ in annual efficiency gains.

    Read case study

  • Featured · AI & ML
    Document chunks
    1,400+
    Programmes covered
    11
    Competency axes
    7

    Lyon: NBS Degree Advisor

    Agentic AI degree advisor · NTU course group project

    An agentic chatbot built on GPT and LangChain that selects between tools: RAG search over 1,400+ document chunks, programme comparison, FAQ lookup and handoff to a human advisor.

    Read case study

More projects

  • AI & ML

    <10 min

    Per query, down from days

    GenAI assistant for the Governor of Lampung's office

    • GenAI
    • RAG
    • LLM
  • Data Science

    0.97

    Best R²

    Multi-source sensor fusion for airport zone occupancy prediction

    • XGBoost
    • LightGBM
    • T-GCN
    • +5
  • Automation & ERP

    17,200+

    Staff-min saved / quarter

    Automated audit reports and invoices for 170+ public healthcare institutions

    • Python
    • Pandas
    • Excel Automation
    • +2
  • Automation & ERP

    5

    Outlets digitised

    Full-stack business system for a multi-outlet restaurant chain

    • Odoo 19
    • Python
    • PostgreSQL
    • +4
  • Data Science

    15%+

    Projected revenue uplift

    Transaction analytics and pricing strategy across 5 outlets

    • Python
    • Pandas
    • Data Analytics
    • +2
  • AI & ML

    95%

    F1 score (BERT)

    Undergraduate thesis: contextual vs non-contextual embeddings

    • BERT
    • LSTM
    • TensorFlow
    • +5

// skills

Skills and capabilities

I work across the full delivery cycle, from defining the business problem to deploying models, pipelines and dashboards.

Machine learning & forecasting

Propensity, forecasting and risk models, from feature engineering through deployment, with SHAP-based explanations for business users.

1M+ customers scored · 3.2% MAPE on weekly price forecasts

  • Python
  • scikit-learn
  • XGBoost
  • LightGBM
  • PyTorch
  • SHAP
  • SageMaker

GenAI & LLM systems

Retrieval-augmented assistants and agentic chatbots grounded in source documents, with escalation to human advisors where required.

1,400+ document chunks indexed · policy answers in under 10 minutes

  • LangChain
  • GPT
  • RAG
  • pgvector
  • Hugging Face
  • BERT
  • FastAPI

Data engineering & automation

Pipelines and workflow automation that replace manual reporting, including ETL into SAP, scheduled Airflow jobs and Power Automate flows.

16 h → 5 min invoice processing · 6 FTE of work automated

  • SQL
  • Snowflake
  • Airflow
  • PostgreSQL
  • Pandas
  • Docker
  • Power Automate

Analytics, BI & experimentation

Dashboards, A/B testing and pricing analysis to support management decisions.

200% revenue growth in 7 months · 15%+ projected pricing uplift

  • Qlik Sense
  • Tableau
  • Power BI
  • Alteryx
  • Streamlit
  • R
  • A/B testing

Full-stack & ERP

Web applications, internal tools and Odoo ERP modules, delivered alongside the underlying data work.

6+ production sites · Odoo ERP across 5 outlets

  • Next.js
  • React
  • TypeScript
  • Tailwind CSS
  • Odoo
  • OWL/JS

// education

Education

2025 – 2026

M.S. Business Analytics

Nanyang Technological University

Singapore

  • CGPA 4.73 / 5.00
  • Dean's List
  • ASEAN Development Scholarship
2018 – 2022

B.S. Computer Science

Universitas Gadjah Mada

Indonesia

  • Cum laude
  • Digital Talent Scholarship

Digital Talent Scholarship awarded by Indonesia's Ministry of Communication and Informatics (Kominfo).

Aug – Dec 2022

Data Science and Machine Learning Program

MIT Institute for Data, Systems, and Society

Online, with Great Learning

    View certificate

    // publications

    Publications

    Peer-reviewed research in natural language processing and health informatics, published during my undergraduate studies at Universitas Gadjah Mada.

    3 papers58+ citationsGoogle Scholar profile
    1. Journal article202255 citations

      A multiclass classification model combining IndoBERT embeddings and long short-term memory for Indonesian-language tweets

      Original title (Indonesian): Model klasifikasi berbasis multiclass classification dengan kombinasi IndoBERT embedding dan long short-term memory untuk tweet berbahasa Indonesia

      Putra, T. I. Z. M., Suprapto, S., & Bukhori, A. F. (2022). Jurnal Ilmu Siber dan Teknologi Digital, Vol. 1, No. 1, pp. 1–28.

      Combined IndoBERT embeddings with an LSTM classifier for Indonesian tweets, reaching a 98.90% F1-score and outperforming Word2Vec-LSTM and fine-tuned IndoBERT baselines.

      doi.org/10.35912/jisted.v1i1.1509
    2. Conference paper20222 citations

      Deep learning and machine learning model comparison for diagnosis detection from medical records

      Heryawan, L., Febriansyah, F., & Bukhori, A. F. (2022). Proceedings of the 2022 International Conference on Computer, Control, Informatics and Its Applications (IC3INA), ACM, pp. 306–310.

      Compared machine learning and deep learning models for detecting diagnoses in physicians' written medical records, a step towards automated ICD coding.

      doi.org/10.1145/3575882.3575941
    3. Conference paper20221 citation

      Comparison of machine learning and deep learning models for Medical Subject Headings indexation

      Heryawan, L., & Bukhori, A. F. (2022). Proceedings of the 2022 International Conference on Computer, Control, Informatics and Its Applications (IC3INA), ACM, pp. 292–295.

      Evaluated logistic regression against an LSTM model for assigning MeSH keywords to biomedical abstracts; the LSTM achieved a higher F1-score on all nine most frequent tags.

      doi.org/10.1145/3575882.3575938

    // contact

    Get in touch.

    I welcome conversations about data science, machine learning and GenAI, including collaborations and speaking opportunities.

    Email me