Skip to content

AI Practitioner & Educator

Kadek Dwi Pradnyana

Helping organizations understand, adopt, and build with AI.

I work with companies across Indonesia to run AI training workshops, build production-ready AI systems, and develop strategy for teams at the start of their AI journey.

Currently at Amartha, building AI for financial inclusion. AWS Certified AI Practitioner.

What I Do

Three ways to work together

01 Popular

AI Training & Workshops

Hands-on workshops for everyone from executives to engineers. Topics include GenAI fundamentals, LLMs in practice, n8n automation, and AI tools adoption. Available on-site or remote.

02

AI Development & Implementation

End-to-end AI systems built for production: chatbots, voice agents, ML pipelines, and credit scoring models. Built to handle real data, real users, and real consequences.

03

AI Strategy & Consulting

Capability assessment, use-case prioritization, vendor evaluation, and AI roadmap development. For organizations at the start of their AI journey who need a clear path forward.

Speaking & Training

Organizations I've trained

Bank Indonesia

Central Bank — Jakarta

Nov 2025

Digital Innovation & AI Workshop

Read the write-up →

Ecommurz

E-commerce Community — Online

Dec 2025

n8n Automation Workshop

Read the write-up →

Bank Lestari

Regional Bank — Jakarta

2025

Generative AI for Banking Teams

Tunaiku (Amar Bank)

Fintech Lender — Jakarta

2025

AI for Credit & Collections Teams

AI Projects Built

Production systems and experiments with measurable outcomes

LangGraph OpenRouter WhatsApp PostgreSQL

AI Telecollection System (Maya)

Improved 7-day payment velocity by 23.5%

WhatsApp AI collection assistant using LangGraph/ReAct pattern. Handles Indonesian language, vision AI for payment receipts, dynamic tone by borrower situation, and human escalation with full context. A/B tested at scale.

  • +23.5% 7-day payment velocity
  • 92.7% conversations resolved without human
  • 22,000+ users in A/B test
Read the full write-up →
AutoGluon Claude Code Python AWS

Credit Scoring Model Pipeline

Full ML pipeline iteration in under 1 hour

Credit scoring pipeline for fintech borrowers using AutoGluon AutoML. Reduced iteration time from ~8 hrs to under 1 hr. Multi-algorithm benchmarking, AUC comparison, and fairness evaluation for thin-file borrowers.

  • Multi-algorithm benchmarking automated
  • AUC comparison across model types
  • Iteration time cut from ~8 hrs to <1 hr
Read the full write-up →

From the Blog

Writing about what I'm building and learning

Ready to bring AI to your organization?

Whether you're looking to upskill your team, build a specific AI system, or figure out your AI strategy — let's talk about what makes sense for your context.