Qualifications / Continuing practice
14 credentials and recognitions across data systems, machine learning, and AI. What matters in my work is how the learning changes a design decision, an evaluation, or a review.
Read my learning notes14
Credentials & recognitions
04
Issuing organizations
Data → ML
Technical foundation
Ongoing
Learning practice
Amazon Web Services · Associate
.png&w=256&q=75&dpl=dpl_BXdF7saZWgFxcu9ifgTJxx45UzEM)
Designing pipelines that survive contact with real data — partition and file layout in S3, orchestration across Glue and Step Functions, and the honest cost/latency tradeoff between batch and streaming. The habit that stuck: model the access pattern before the schema, and make every stage idempotent so a replay is boring.
Amazon Web Services · Associate
The distance between a model in a notebook and a model under load — packaging, endpoints, autoscaling, and the drift you only ever notice in production. It taught me to treat the model as the smallest part of the system, and the pipeline around it as the real deliverable.
Amazon Web Services · Specialty
.png&w=256&q=75&dpl=dpl_BXdF7saZWgFxcu9ifgTJxx45UzEM)
Deeper into the modeling side: feature engineering, algorithm selection, tuning, and the evaluation discipline that separates a useful metric from a flattering one. Where I learned to distrust a single accuracy number and design evaluation around the decision the model actually supports.
Amazon Web Services · Professional
.png&w=256&q=75&dpl=dpl_BXdF7saZWgFxcu9ifgTJxx45UzEM)
Authoring and defending professional-level exam items means every option has to be justified against documentation — no appeals to taste. It rewired how I review architecture: state the constraint, justify the tradeoff, and cut anything that can't be sourced.
Amazon Web Services · Associate
.png&w=256&q=75&dpl=dpl_BXdF7saZWgFxcu9ifgTJxx45UzEM)
Calibrating what associate-level competence actually means across thousands of candidates. The discipline is precision of language — testing whether someone understands a concept rather than whether they memorized a service page.
AWS Microcredential · Demonstrated
Agent design where it counts: tool schemas, planning loops, memory boundaries, and the guardrails that stop an autonomous system from confidently doing the wrong thing. This feeds directly into Praxis — agents that have to show their evidence before they are allowed to act.
AWS Microcredential · Demonstrated
CI/CD for models — versioned data, reproducible training runs, automated evaluation gates, and a rollback that still works at 2am. The core lesson is blunt: if you can't reproduce it, you don't own it.
Accenture
The enterprise view of where agentic systems create value in a large organization — and where they quietly fail: process fit, human-in-the-loop design, and change management. A useful counterweight to the engineering instinct to automate first and ask later.
Anthropic · Foundations
Architecting on Claude: context engineering, tool use, prompt caching, and the real economics of long-context systems. It sharpened how I scope LLM work — pick the model for the decision, keep the context auditable, and measure grounding rather than fluency.
Palantir
Foundry's ontology-first model — pipelines, datasets, and transforms bound to an object model instead of loose tables. It changed how I think about lineage: when every transform is tracked and every object carries provenance, governance stops being a document and becomes part of the build.
Amazon Web Services · Professional
Cloud architecture across security, reliability, migration, and operations. The architectural questions connect infrastructure decisions to the constraints of the organization running it.
Amazon Web Services · Business
AI strategy, use-case selection, and organizational readiness. A useful connection between technical feasibility and the business decision an AI system is expected to improve.
Amazon Web Services · Recognition
Early Adopter recognition associated with the AWS Certified AI Business Strategist credential. Recorded here alongside the certification as a companion recognition.
Amazon Web Services · Professional
Building generative AI applications with attention to retrieval, evaluation, security, and operational reliability. These concerns connect application development to the evidence needed to trust its outputs.
Learning into practice