Ten Very Hard Benchmark Tasks in Ten Days
First AI benchmark job: ten Indonesian tasks in ten days, half merged, most rated very hard, the top band. Over 300 ideas killed. The method, not the tasks.
Writing
My writing, research publications, published models and teaching materials.
Blog
Oct 27, 2026
Moving ARIA From ACS to LiveKit
ARIA moved from ACS to self-hosted LiveKit. A PoC drill measured 425 ms median hand-back over 20 rounds; no matched ACS-to-LiveKit speedup was measured.
Oct 24, 2026
Untrusted Code Next to Production Secrets
An audit showed a repository's code could run inside the API holding production secrets. Same-day stopgap, then isolated workers and a 9-minute outage.
Oct 24, 2026
The Stall That Was Never in the Query
A Fabric dashboard took ~21 s per request, static files included. A zero-row probe showed capacity throttling, not query cost. I had blamed indexes first.
Oct 24, 2026
The Install Lock That Also Blocked the App
In the first managed application test installs, every AI call failed with correct roles. The lock blocked the app's own identities. 36 defects followed.
Oct 16, 2026
Clearbox: Inbox Triage with Microsoft-Decision-1
Clearbox pre-sorts Gmail and Outlook email for approval, using Microsoft-Decision-1 and guardrails. The evaluation covered 40 synthetic emails, not an inbox.
Oct 9, 2026
How We Made a Voice Agent Interruptible
Voice-agent barge-in reached ~0.2 s in a controlled knowledge-search harness, n=4 calls per arm. Later live cut-off tests showed no improvement.
Oct 9, 2026
Why the Better Fraud Engine Shipped in Shadow Mode
In offline tests on the noisy-clean development slice, FPR fell 36.4% to 9.1%, but High-risk recall fell. The engine shipped only in shadow mode.
Oct 9, 2026
45% of Our LLM Calls Failed: What the Error Strings Told Us
About 45% of 22 CV/job pairs failed. Two parsing bugs were fixed with seven regression tests, but no rerun of the original batch was recorded.
Oct 9, 2026
Spec-Driven Development with AI Coding Agents
AI agents wrote most of a solo-built vendor portal. CI missed a skipped integration tier and two sign-in bugs. Productivity gains were not measured.
Jun 6, 2025
The Sun, The Moon, and The Dark Sea
A personal essay on growing up, finding direction and staying steady, written in Bahasa Indonesia.
Feb 28, 2022
CRUD in Node.js with the Express Framework and MySQL (opens in a new tab)
A beginner-friendly tutorial for building a CRUD REST API with Express and MySQL (in Bahasa Indonesia).
Dec 31, 2020
UI/UX Design Case Study: A Baby Development Tracking App (opens in a new tab)
How we designed Helby, an app that helps parents track their baby's development, from research to prototype (in Bahasa Indonesia).
Research
My peer-reviewed papers and BSc thesis publication.
13
2
13 citations · h-index 2 on Google Scholar, as of Oct 2026
Implementasi Model IndoBERT pada Dashboard Sentimen Media Sosial (Studi Kasus Universitas XYZ)
Authors: Kurniadi Ahmad Wijaya
eProceedings of Engineering, Vol. 10 No. 4, Telkom University
Tracks social media sentiment about a university in an IndoBERT dashboard (BSc thesis publication).
Sentiboard: University Sentiment Dashboard Using IndoBERT
Authors: Kurniadi Ahmad Wijaya
Telkom University
My BSc thesis: Sentiboard, a university sentiment dashboard using IndoBERT.
Study on the Dynamics of COVID-19 Cases in Achieving Herd Immunity in Indonesia
Authors: Pratama Azmi Atmajaya, Kurniadi Ahmad Wijaya, Alif Adwitiya Pratama, Aditya Firman Ihsan
2021 International Conference on Intelligent Cybernetics Technology & Applications (ICICyTA), IEEE
A historical 2021 study using COVID-19 case forecasts to examine Indonesia's path to herd immunity.
DOI 10.1109/icicyta53712.2021.9689122 (opens in a new tab)Source code (opens in a new tab)
Indonesian Digital Wallet Sentiment Analysis Using CNN and LSTM Method
Authors: Ananda Affan Fattahila, Fendi Irfan Amorokhman, Kaenova Mahendra Auditama, Kurniadi Ahmad Wijaya, Ade Romadhony
2021 International Conference on Artificial Intelligence and Big Data Analytics (ICAIBDA), IEEE
Analyzes sentiment in Indonesian digital wallet app reviews with CNN and LSTM.
Read the paperDOI 10.1109/icaibda53487.2021.9689712 (opens in a new tab)Source code (opens in a new tab)
Models and datasets
2024
IndoBERTweet App-Review Sentiment Model (opens in a new tab)
Model, Hugging Face
2024
IndoBERT App-Review Sentiment Model (opens in a new tab)
Model, Hugging Face
2022
Talks and workshops
Oct 2026
From a Request to a Tool-Using Agent with Tracing in Microsoft Agent Framework
Developer workshop presentation: Microsoft Agent Framework in Python
Sep 2026
Microsoft AI for Business Workflows
AI workshop for an industrial enterprise client (presentation author)
Mar 2022
Study Groups for Basic Artificial Intelligence
Telkom University (Lecturer Assistant)
Jan 2022
Hackathon Judging and Team Mentoring
Motion Laboratory hackathon, Telkom University
Sep 2020
Go Programming Fundamentals Practicum
Telkom University (Practicum Assistant)
Jan 2020
Workshops and Competitions for the Programming Club
Proclub Telkom University
Study guides
The study guides written while preparing each certification exam, readable right here.
Developing in Agentic AI Systems
Operating, supervising, evaluating and governing AI agents across the software development lifecycle, with GitHub as the system of record and control plane.
- Agent architecture and SDLC 15-20%
- Tool use and environment 20-25%
- Memory, state and execution 10-15%
- Evaluation and tuning 15-20%
- Multi-agent coordination 15-20%
- Guardrails and accountability 10-15%
GitHub Copilot
Using GitHub Copilot responsibly and productively: plans and features, Copilot Chat, agent mode and the CLI, data flow, prompt engineering, testing and the privacy controls administrators configure.
- Use Copilot responsibly 15-20%
- Use Copilot features 25-30%
- Data and architecture 10-15%
- Prompt engineering and context 10-15%
- Developer productivity 10-15%
- Privacy, exclusions and safeguards 10-15%
Generative AI Leader
Gen AI fundamentals, Google Cloud's gen AI offerings, techniques to improve model output and the business strategy for secure, responsible adoption.
- Fundamentals of gen AI ~30%
- Google Cloud's gen AI offerings ~35%
- Techniques to improve model output ~20%
- Business strategy for gen AI ~15%
Agentic AI Business Solutions Architect
Planning, designing and deploying agentic AI business solutions across Microsoft 365 Copilot, Copilot Studio, Microsoft Foundry, Power Platform and Dynamics 365, with architecture case studies and original practice questions.
- Plan AI-powered business solutions 25-30%
- Design AI-powered business solutions 25-30%
- Deploy AI-powered business solutions 40-45%
Developing AI Apps and Agents on Azure
Microsoft Foundry, agents, vision, text analysis and information extraction, with decision tables, verified code and original practice questions.
- Plan and manage an Azure AI solution 25-30%
- Generative AI and agentic solutions 30-35%
- Computer vision solutions 10-15%
- Text analysis solutions 10-15%
- Information extraction solutions 10-15%
Microsoft Azure AI Fundamentals
A beginner-friendly path through responsible AI, how generative models work, AI workloads and hands-on Python with Microsoft Foundry.
- AI concepts and capabilities 40-45%
- AI solutions with Microsoft Foundry 55-60%
Independent study material, not an official exam guide.