[Telkom University] The First Step into the Universe of Insight - PKKMB S1 Sains Data 2025

"The First Step into the Universe of Insight" takes audiences on a journey through the world of Data Science, where persistence, adaptability, and critical thinking transform raw information into breakthrough innovations. The presentation shares the speaker's seven-year evolution from Statistics to Data & AI, including the challenges that built expertise across multiple industries. It explores the data ecosystem's key players, from Data Engineers to AI Product Managers and demonstrates how these roles collaborate to create real business value. Through stories from finance, e-commerce, healthcare, manufacturing, and retail, attendees discover how Data Science drives everything from catching fraud to personalizing customer experiences, predicting equipment failures, and advancing precision medicine. The session pulls back the curtain on how Data Science actually works, the iterative cycle of defining problems, preparing data, building models, and deploying solutions. It also addresses what it takes to succeed in this competitive field: continuous learning, finding community, and staying curious. The presentation concludes with an important truth: Data Science isn't something mastered overnight it's a lifelong journey where those who embrace AI and innovation will shape tomorrow, while those who don't risk being left behind.

Transcript

1. The First Step into the Universe of Insight PKKMB S1 Sains Data 2025 Telkom University | Thursday, 2025-09-11 Copyright@2025 Paper.id - Confidential & Proprietary Webinar Series TGIF #1
2. Data & AI executive with 7+ years of experience in Data Science, Analytics, and Engineering, currently leading AI and data innovation at Paper.id. He has developed a diverse skill set across sectors such as financial technology, management consulting, e-commerce and beyond. Beyond his professional role, he has delivered over 25+ both national and international talks and has served as an Advisory Board Member for the Statistics Department at Universitas Padjadjaran, an Adjunct Assistant Lecturer of a Master Degree of Faculty of Economy and Business at Universitas Indonesia, a Guest Lecturer at the School of Business and Management, Institut Teknologi Bandung, and many more. Fiqry Revadiansyah Div. Lead Data Science and Engineering Speakership Portfolio Copyright@2025 Paper.id - Confidential & Proprietary Webinar Series TGIF #1
3. Data Science Personal Journey - 5’ 01 02 Data Science in Industry - 5’ 03 Data Science Methodology - 5’ The First Step into the Universe of Insight 04 Adaptation and Fellowship - 5’ 05 Closing Remark - 5’ Copyright@2025 Paper.id - Confidential & Proprietary Webinar Series TGIF #1
4. Data Science Personal Journey Copyright@2025 Paper.id - Confidential & Proprietary Webinar Series TGIF #1
5. Data Science Personal Journey An unpredicted privilege from Statistics to Data & AI, journey of 7 years of challenges. 1. Journey 4 5 2 3 Copyright@2025 Paper.id - Confidential & Proprietary The First Step into the Universe of Insight Data Science Personal Journey Webinar Series TGIF #1 05-2018 Graduated with a Bachelor’s degree in Statistics. Thesis: “Generalized Space Time Autoregressive Model,” using R programming language. 05-2018 Took internship and freelance jobs as a Data Scientist. Passed entry test using Python with a 1-week deadline, quickly learning all required skills within the timeframe. 09-2018 First full-time Data Scientist role, focused on customer growth, fraud prevention, and chat analytics. 05-2021 Worked as a consultant for Data Science projects in telecom projects. Built ML pipelines, anomaly detection, and time-series models to speed up deployments. 03-2022 Senior Data Scientist at Indonesia’s unicorn investment platform. Led analytics migration and predictive modeling, and analytics engineering. 07-2023 First time managerial position, Division Lead of Data Science & Engineering. Managed teams (DS, DE, AI, AI PM, AI Ops), built AI-native products and data engineering pipelines and architecture. Passionate in Academic Passionate in Learning by Sharing Now
6. Data Science Personal Journey Data Science is not an overnight achievement, it is shaped through rigorous thinking, persistence, constant refinement and iteration. 1. Journey 4 5 2 3 Copyright@2025 Paper.id - Confidential & Proprietary The First Step into the Universe of Insight Data Science Personal Journey Webinar Series TGIF #1 Leadership in Data and AI Transforming vision into execution, bridging AI innovation, enterprise data platforms, and organization leadership at scale. Specialist to Generalist Sharpening specialities while expanding perspective through cross-functional data role collaboration, adapting to evolving technologies, and maintaining consistent quality of work. A New Lens of Understanding A turning point in business understanding revealed that Data Science is not only about ML products, but also about measuring impact and blending with product-market fit theory. A Guess That Changed Everything A first opportunity in the Data Science field, requiring a complete shift from programming languages to technical data theory, pursued with full dedication and hope. Strengthening the Foundation Passionate about coding, enriched by a statistical mindset, and specialized in Time Series Modeling, which gave the confidence to pursue a career in Data Science. Learning Journey of Data Science Reality in Venn Diagram
7. Data Science in Industry Copyright@2025 Paper.id - Confidential & Proprietary Webinar Series TGIF #1
8. A Glimpse Day with Data Science Without realizing it, the moment we wake up until we go back to sleep, Data Science nearby us 4 5 3 Copyright@2025 Paper.id - Confidential & Proprietary The First Step into the Universe of Insight Data Science in Industry Webinar Series TGIF #1 1 2. Industry
9. Data Roles in the Industry Mapping the journey of data professionals, from building pipelines and deploying AI models to driving insights and managing AI-powered products. 4 5 3 Copyright@2025 Paper.id - Confidential & Proprietary The First Step into the Universe of Insight Data Science in Industry Webinar Series TGIF #1 1 2. Industry Building the Data Foundation Build data pipelines, ensure data quality, and manage infrastructure. Core Tools: SQL, Apache Kafka, dbt, Airflow, Spark, Databricks, Snowflake, BigQuery/Redshift, etc. Data Engineer 1 ML/AI Engineer 2 3 Analytics Engineer 4 Data Scientist 5 Data Analyst 6 AI Product Manager Turning Models into Reality Develop, train, and deploy machine learning/AI models at scale. Core Tools: Python, TensorFlow, PyTorch, scikit-learn, Vertex AI, MLflow, Docker/Kubernetes Bridging Data and Insights Transform raw data into analysis-ready datasets and maintain analytics workflows. Core Tools: dbt, SQL, Looker, Power BI, Tableau, GitHub From Data to Decisions Analyze data, build predictive models, and generate actionable insights. Core Tools: Python (pandas, NumPy, scikit-learn), R, TensorFlow, Jupyter, Spark, BigQuery, Vertex AI, TensorFlow, etc. Telling Stories with Data Query data, create dashboards, and translate data into business recommendations. Core Tools: Excel, SQL, Tableau, Power BI, Looker Data Studio Driving AI-Powered Products Define vision and strategy for AI products, align technical outputs with business needs. Core Tools: Jira, Confluence, Figma, Vertex AI More Engineering More Strategic
10. Data Implementation in the Industry How Data Science drives automation, intelligence, and innovation in today’s industries. 4 5 3 Copyright@2025 Paper.id - Confidential & Proprietary The First Step into the Universe of Insight Data Science in Industry Webinar Series TGIF #1 1 2. Industry Finance e-Commerce Healthcare Manufacturing Retail Fraud detection & credit scoring: ML models (gradient boosting, graph neural networks) analyze transactions and behavior to flag anomalies and assess credit risk. OCR for KYC/KYB: Computer vision + NLP extract, or using VLM (Visual Language Model) data from identity documents, enabling instant verification. Recommendation system: Content-based filtering and collaborative filtering personalize product discovery. Dynamic pricing: Reinforcement learning and demand prediction adjust prices based on inventory, competition, and user behavior. Customer segmentation: Clustering (k-means, DBSCAN) groups customers by demographics, purchase history, and engagement. Diagnosis support: Deep learning (CNNs) on X-rays, MRIs, CT scans detects early signs of disease. Medical image analysis: AI segmentation and classification identify tumors, lesions, and anomalies with high precision. Precision treatment: Predictive models on genetics, EHR, and wearables enable personalized treatment plans. Predictive maintenance: Time-series models and anomaly detection on IoT sensor data forecast equipment failures. Supply chain optimization: AI forecasting and optimization improve inventory, logistics routing, and supplier risk management. Digital twins: Virtual replicas simulate machines/plants for performance testing and cost optimization. Demand forecasting: ML models (XGBoost, LSTM) integrate sales history, promotions, and external data to predict demand. Personalization: Generative AI creates tailored product descriptions, recommendations, and marketing visuals in real time.
11. Data Implementation in the Industry How Data Science drives automation, intelligence, and innovation in today’s industries. 4 5 3 Copyright@2025 Paper.id - Confidential & Proprietary The First Step into the Universe of Insight Data Science in Industry Webinar Series TGIF #1 1 2. Industry Finance Fraud detection & credit scoring: ML models (gradient boosting, graph neural networks) analyze transactions and behavior to flag anomalies and assess credit risk. OCR for KYC/KYB: Computer vision + NLP extract, or using VLM (Visual Language Model) data from identity documents, enabling instant verification. Payment Reconciliation User Verification - KYC Fact 1 receipt = 30 seconds of manual process (typing, checking, context understanding, etc.) Using ML/AI models, this could be done under 3 seconds/image* The same applies to user KYC, beyond speeding up the process, Intelligent Document Processing (IDP) also helps verify and clarify whether an image is genuine or fraudulent.
12. Data Science Methodology Copyright@2025 Paper.id - Confidential & Proprietary Webinar Series TGIF #1
13. Data Science Methodology Framework Interdisciplinary, iterative, and insight-driven, a lifecycle that fuses multiple expertises Copyright@2025 Paper.id - Confidential & Proprietary The First Step into the Universe of Insight Data Science Methodology Webinar Series TGIF #1 1 2 3 4 Analogy in a “Restaurant” *Foundational Methodology for Data Science (IBM, 2015) 1 A customer walks in with a request. The chef must first understand what the diner really wants and plan the recipe. Listen to the Inquiries 2 Listing the requirements, going to the market, and selecting fresh ingredients is like collecting/understanding the data. Pick the Ingredients 3 Ingredients must be cleaned, chopped, and prepared before cooking, where it needs to be structured, tested, and clean. Getting the Pieces Ready 4 The meal is served to the customer. Feedback, reviews, and ratings decide whether the dish is needs improvement. Trying It Out in Real Life 4 5 1 3. Method 2
14. Data Science Methodology Framework How Data Teams Transform Transaction Spikes into Actionable Fraud Insights 4 5 Copyright@2025 Paper.id - Confidential & Proprietary The First Step into the Universe of Insight Data Science Methodology Webinar Series TGIF #1 1 3. Method 2 Question from Bu Sopi Finance Operation Manager – Finance Division "Minggu lalu ada apa ya? Kok bisa budget promo kita tinggal segini lagi? Tolong cek dong ini ada error sistem atau apa?” Question from Mba Tasya Product Manager – Onboarding Department "Tim marketing lagi rilis apa ini, kok tiba-tiba banyak banget user yang register dan convert transaksi pertama? Mana nama username nya aneh, kayak fufafufa, mr.barusadar, hidupJKL” Data Team Activity Fraud Transaction Case Alomani transaksi Problem Understanding & Scoping Identify the unusual transaction spike and define the fraud problem clearly. • Confirm with finance and marketing whether the spike is expected (e.g., promo campaign) or abnormal. • Translate the business concern into a data problem: “Classify transactions as normal or fraudulent.” 2 Data Collection & Exploration Gather all relevant transaction and user activity data. • Transactions: time, amount, payment method, promo code usage. • Users: device, IP address, location, account age. • System logs: error codes, retry attempts, anomalies. 3 Data Preparation & Feature Engineering Clean and transform the raw data for analysis. • Remove duplicates, fix missing values. • Create fraud indicators (e.g., multiple accounts using same promo code, same IP but different cards). • Balance the dataset (fraud vs normal) to avoid biased models. 4 Modeling & Validation Build and test machine learning models for fraud detection. • Classification models (e.g., Logistic Regression, Random Forest, Gradient Boosting). • Evaluate with metrics like Precision, Recall, F1, ROC-AUC to handle imbalanced data. 1 5 Deployment & Actionable Insights Put the fraud detection system into production and enable business action. • Real-time classification: flag or block suspicious transactions instantly. • Send alerts to risk/finance teams for manual review. • Provide dashboards and reports for executives to track fraud patterns and campaign health.
15. Adaptation and Fellowship Copyright@2025 Paper.id - Confidential & Proprietary Webinar Series TGIF #1
16. What It Really Takes to Land Your First Role in Data A story of hiring experience for Internship position at Paper.id 5 Copyright@2025 Paper.id - Confidential & Proprietary The First Step into the Universe of Insight Adaptation and Fellowship Webinar Series TGIF #1 1 4. Adapt 3 2 Acceptance Rate acceptance rate for DSE (Data Science, AI and Engineer) Internship 0.0025% (1 / 400) Any many more… Application Data Reveal Is the data legit? Here’s what the ‘competitor’ looks like:
17. What It Really Takes to Land Your First Role in Data A story of hiring experience for Internship position at Paper.id 5 Copyright@2025 Paper.id - Confidential & Proprietary The First Step into the Universe of Insight Adaptation and Fellowship Webinar Series TGIF #1 1 4. Adapt 3 2 Previous Accepted Candidates What most recruiters care about 1 Has multidisciplinary interests and a deep understanding of the role, completing multiple certifications and classes. Highly Passionate Person 2 Relevant experience that aligns with the position, highlighted by 0→1 projects that show initiative and end-to-end execution. Internship/Portfolio Projects 3 Strong drive to explore new tools and concepts, with the ability to adjust quickly and thrive in changing environments. High Curiosity and Adaptive
18. Exploring Pathways to Learn Data Science From platforms to communities, resources are everywhere to support learning 5 Copyright@2025 Paper.id - Confidential & Proprietary The First Step into the Universe of Insight Adaptation and Fellowship Webinar Series TGIF #1 1 4. Adapt 3 2 Community-driven Platform LMS-driven Platform • Jack of all trades of Data Science-related, biggest community in Indonesia • Multi-channel platform, like WhatsApp group, etc. • Google-based community in Bandung region, very informative, often shares free learning coupon codes (cloud skill boost) for google cloud certification / upskill. • Python-based community in Indonesia, sometimes shares DS/AI/ML-related topics, or publish events related to python use case in Industry.. • Top tier learning platform for university students, learn the theory with the real industrial use cases, taught by industry experts from Google, GoTo, and Traveloka engineers. • Or, try to learn from these online coding exercises and learning platform: • From Data Science to Software Engineering in Bahasa Indonesia • The most up-to-date AI/DS technology introduction and tutorial
19. Surviving the First Break: How to Land a Data Science Every step is a battle to stand out in one of the most competitive fields today 5 Copyright@2025 Paper.id - Confidential & Proprietary The First Step into the Universe of Insight Adaptation and Fellowship Webinar Series TGIF #1 1 4. Adapt 3 2 ☐ Learn fundamental programming (Python, R, SQL) ☐ Understand basic statistics & linear algebra ☐Start introductory courses (Coursera/Datacamp/LeetCode basics) ☐ Join DS/AI communities (Data Science Indonesia, GDG, PythonID) ☐ Build simple mini projects (EDA public datasets, basic visualization) ☐ Choose a specialization track (NLP, CV, or classical ML) ☐ Master industry tools (Pandas, scikit-learn, XGboost, TensorFlow, PyTorch) ☐ Earn certifications (AWS/GCP ML, dbt, SQL) ☐ Start real portfolio projects (recommendation system, fraud detection, sentiment analysis) ☐ Learn prompt engineering & AI tools (LangChain, workflow automation, vibes coding) ☐ Join internships or part-time data jobs ☐ Build an end-to-end capstone project (business problem → data pipeline → model → dashboard) ☐Try online competitions for portfolio (Kaggle, Hackathon) ☐ Find mentor (Adplist.org, Data Science Indonesia, etc.) ☐ Complete a portfolio on GitHub / personal website ☐ Write projects in business language (“reduced fraud rate by 15%”) ☐ Prepare a measurable CV & LinkedIn profile ☐ Practice coding tests and mock interviews ☐ Contribute to communities (mentoring, open-source) Year 1–2 (Freshman / Early Stage) Year 2–3 (Middle Stage) Year 3–4 (Pre-Grad Stage) Final Year (Graduate Readiness) Tools to Explore AI Trends in 2024 - 2027 Coding Platform LLM Tools Vibes Coding Interpretable & Explainable AI Generative AI Model Context Protocol (MCP) Agentic AI Small Language Model (SLM) Context Engineering
20. Closing Remark Copyright@2025 Paper.id - Confidential & Proprietary Webinar Series TGIF #1
21. Orchestrating the Data Science Journey Never settle for less, Data Science is not an overnight journey Copyright@2025 Paper.id - Confidential & Proprietary The First Step into the Universe of Insight Closing Remark Webinar Series TGIF #1 1 2 3 5. Closing 4 Math & Stat ML & Deep Learning NLP, CV, RL Generative AI - IPK 4.0 - Juara hackathon internasional 10x - Peserta COC - Lolos coding test di MAANG Minta ChatGPT yg ngerjain Tanpa nyari tau “how-to”nya
22. The AI Revolution Waits for No One Moore’s Law: Thriving in an Era Where AI Capabilities Double Every Few Months Copyright@2025 Paper.id - Confidential & Proprietary The First Step into the Universe of Insight Closing Remark Webinar Series TGIF #1 1 2 3 5. Closing 4
23. Data Science Closing Statement Survival starts with action; no time for regret when you are already left behind Copyright@2025 Paper.id - Confidential & Proprietary The First Step into the Universe of Insight Closing Remark Webinar Series TGIF #1 1 2 3 5. Closing 4 AI won’t replace people. But those who don’t use AI will be replaced
24. THANK YOU Fiqry Revadiansyah Division Lead of Data Science and Engineering @Fiqry Webinar Series TGIF #1