AI & Machine Learning
Classification, forecasting and anomaly detection — evaluated on real operational data rather than curated benchmarks.
- Python
- ML workflows
- Model evaluation
I build machine learning, analytics and business intelligence systems on data that rarely arrives clean — grounded in three years of telecommunications operations, provisioning validation and ISO 27001 security work.
Classification, forecasting and anomaly detection — evaluated on real operational data rather than curated benchmarks.
Turning operational records into dashboards decision-makers actually open — and keeping them trustworthy once they are in use.
Validation, reconciliation and end-to-end quality control across high-volume provisioning data — where most model failures actually originate.
ISO 27001 Lead Implementer with hands-on certification and audit experience — an unusual pairing with AI work in regulated industries.
Operator-side domain knowledge: provisioning flows, service activation chains, acceptance testing and vendor coordination on live infrastructure.
Alibaba Cloud certified in computing and security — the infrastructure grounding to deploy data and ML workloads, not only prototype them.
Operational data work at telecommunications scale, information security programmes, and the applied machine learning portfolio I'm building alongside my master's.
Notes on machine learning that has to survive contact with real operational data — validation, governance, and the unglamorous parts that decide whether a model works.
I'm Ilham Dyki Mu'ahfi, an AI Engineer based in Jakarta. My route into machine learning went through telecommunications operations rather than research — three years of project validation, provisioning data and security consulting before formalizing the AI side academically.
The model is the easy part. The data describing reality accurately is the hard part.
That belief comes from experience rather than principle. At Telkomsel I handled a document archive where the same technical fact could be recorded three different ways. At Telkom Akses I worked five hundred cases where service activation failed because two systems disagreed about a customer. In both, the interesting problem was never the algorithm — it was establishing what was actually true.
The consulting year at Aktuator Cipta Cendekia added something different. Implementing ISO 27001 for enterprise clients meant learning how organizations adopt controls, and why technically correct solutions fail when nobody's trained on them. As AI moves into regulated industries, that turns out to be unusually relevant.
I'm now completing a Master of Informatics Engineering at BINUS University with an artificial intelligence focus, alongside a data analytics bootcamp at Dibimbing — pairing formal machine learning theory with operational grounding. I'm looking for AI engineering and data roles where both halves matter.