Credentials
28 verified credentials, earned alongside the projects they fed into. Each one links to the original certificate.
Covers how generative models differ from traditional ML, where they fit in the Google Cloud stack, and which problems they are a poor fit for.
How LLMs are trained and tuned, why prompt design changes output quality, and what prompt tuning costs compared to fine-tuning.
Prompt patterns that hold up in production: zero-shot and few-shot, chain of thought, and the interview pattern for pulling detail out of a model.
Generative models across text, image, code, and audio, and the tooling around each. Ends on where the technology is applied in real products.
How foundation models are pre-trained and adapted downstream, and what separates the major model platforms in practice.
Why responsible AI is an engineering problem rather than a policy afterthought, and how Google structures its own AI principles.
Turning AI principles into operational practice: governance structures, review processes, and the tradeoffs teams hit when applying them.
Azure OpenAI Service end to end, from model deployment to prompt design and the responsible AI controls built into the platform.
Language understanding on Azure AI Language: entity and intent recognition, sentiment, translation, and conversational speech.
Image classification, object detection, and face and OCR services on Azure AI Vision, including when a custom model beats a prebuilt one.
Pulling structure out of unstructured documents with Azure AI Document Intelligence, and building searchable indexes with Cognitive Search.
The Azure AI service catalogue and how the pieces fit together, plus the workload types each service is meant to handle.
Agent architectures that plan, call tools, and keep state across steps. The design questions here fed directly into my 6-agent Aura pipeline.
Building a working assistant end to end: grounding it in your own data, wiring up tools, and handling conversation state.
Structured prompting for real applications, including how to constrain output format and reduce hallucination on retrieval tasks.
The mechanics underneath generative models: how they are trained, why they hallucinate, and what that means for building on them.
Where generative AI earns its cost in a business, how to scope a pilot, and the risks worth raising before one ships.
Neural network architectures and the training decisions that matter: optimizers, regularization, and reading a loss curve honestly.
Using AI tooling inside a delivery workflow for planning, estimation, and tracking, without handing over decisions that need a human.
A 12 week proctored course covering supervised and unsupervised methods and model evaluation. Scored 71% and earned the Elite grade.
An 8 week proctored course on algorithm design and complexity analysis. Scored 75%, placing in the top 5% of 1,696 certified candidates.
An 8 week proctored course on core data structures and their Python implementations, assessed by assignments and a written exam.
Server side fundamentals and how the backend layers sit behind a web application, plus where each common framework is a sensible choice.
Python for production work: data structures, object oriented design, error handling, and testing.
Linux administration from the command line: users and permissions, processes, storage, and shell scripting on RHEL.
The second RHEL course, covering networking, storage management, SELinux, and task automation.
How agile teams actually run: iteration planning, estimation, and the practices that survive contact with a real backlog.
Running a board in Jira: configuring workflows, managing a backlog, and reading burndown and velocity reports.