Build across boundaries
Ingestion, retrieval, modeling, evaluation, reporting and interface—not a disconnected proof of concept.
I build AI systems
from source to decision.
I work across data, retrieval, models, evaluation and the operating workflow around them—so the result is observable, defensible and useful to the people making the call.
- Lead AI engineer
- MSc psycholinguistics
- Edmonton, Canada
The whole system is the work.
I build systems that move messy source material into decisions leaders can defend—and I stay close to every handoff where quality, cost or accountability can disappear.
That means being fluent with the engineers building the pipeline and the executives steering the business on what it produces. The interface matters. So do the contracts, tests, observability and review practices underneath it.
Design for evidence
Ground truth, citations, traceability and error analysis built into the system instead of added after launch.
Keep judgment human
Automation accelerates the work; architecture, review and consequential decisions stay deliberately owned.
Selected work
Three examples of technical depth translated into operational outcomes.
Retrieval systems for regulated documents
Multi-tenant retrieval over regulations, policies and procedures, combining dense, sparse and graph retrieval with citations and provenance a compliance team can inspect.
Inspect the retrieval lab
Source → cited answer
retrieval quality measured against labeled evidence
Security intelligence that executives can use
Automated reporting pipelines that turn NVD and MSRC feeds into quarterly and weekly security intelligence, including an edition that analyzed 249 CVEs.
Inspect a real quarterly report
249 CVEs
scored, categorized and translated into an executive artifact
Automation tied directly to the business
Built the ecommerce, learning and marketing automation behind thousands of customer relationships—technology designed around the promise the organization made to its learners.
$3.5M
first-year sales supported by the platform and automation
The proof is interactive.
A working slice of the retrieval system runs in the browser. Change the strategy and watch Recall, MRR and nDCG respond against labeled ground truth.
Open the retrieval labHow I think about the work
Technical judgment is part of the system. These are two places where I have made that practice inspectable.
Technical report · August 2026
Extraction stopping calibration
A probabilistic approach to deciding when another graph RAG extraction pass is unlikely to add enough novel evidence to justify its cost.
Agents write the code. Someone still owns it.
Across 459 tracked proposals, not one was implemented perfectly by a frontier model. The gap between passing tests and belonging in the codebase is where architecture and review live.
- 459 → 0
- tracked proposals → perfect first implementations
- 31%
- of commits were review-driven work
- 21
- blocking rules created from real failure classes
A production stack, with a job for every tool.
The stack sits further down because tools support the work; they are not the story. For a technical reviewer, the full operating surface is here.
Data and orchestration
Dagster Cloud Serverless · PostgreSQL · Supabase
Event-driven pipelines, auditable migrations, cost controls and row-level governance.Retrieval and graph
Haystack AsyncPipeline · Milvus / Zilliz · Qdrant · Neo4j Aura
Hybrid retrieval, reranking, graph traversal, provenance and streamed answers.Machine learning
SetFit · ONNX · cross-encoders · statistical inference
Corpus-specific classifiers, quality gates, experiments and defended evaluation.Cloud and operations
Azure · managed identity · containers · Service Bus
Enterprise authentication, secrets, asynchronous workloads and GPU inference.Agentic engineering
Harnesses · rules · skills · Langfuse · structlog
Supervised implementation, blocking contracts and end-to-end observability.
Edmonton, Alberta
The long way here was the training.
I started in psycholinguistics—an MSc at the University of Alberta, designing experiments on how people process language and defending every inference against a committee. Retrieval quality, embedding behavior and whether an answer follows from its sources are still language-processing questions with a statistical core.
The path between was not a detour. I ran companies, built automated physical systems, shipped an ecommerce and learning platform, retrained deliberately in data science, and moved into security intelligence and applied AI. The through-line is simple: show the evidence, and build the system around the decision it has to support.
Field notes from the bench
Experiments, failures and durable lessons from building production AI systems.
Bring me the problem that crosses the boundaries.
That is usually where the interesting work starts.
