0+
Evals Implemented
Offline and online evaluations deployed.
AI Engineering
Evaluation-first AI systems with model telemetry, guardrails, and trace-driven iteration. Built by Kaan Un, founder of Traction IT.
Shader Layer Active
WebGL background and atmospheric motion load only on this route group.
System Performance
System Architecture
Eval Pulse
0+
Evals Implemented
Offline and online evaluations deployed.
0%
Latency Improvement
Reduction in end-to-end inference latency.
0+
Model Iterations
Controlled experiments completed with trace instrumentation.
0%
Guardrail Coverage
Core user flows covered by safety and quality checks.
Model + Data + Delivery
LLM Evaluation Design
Creates offline and online evals for quality and safety.
Error Analysis
Uses traces to identify failure clusters and root causes.
RAG Systems
Builds retrieval pipelines with observability and guardrails.
Model Serving
Deploys low-latency endpoints with fallback strategies.
Telemetry
Instruments LLM workflows for debugging and monitoring.
Synthetic Data
Generates focused test sets for edge cases and regressions.
Latest Publications

2026-03-14
The real alpha is not in gradient apps or vibe-coded clones. It is in accelerated data enrichment and AI-native quantitative research.

2026-03-09
I burned through my remaining credits to deploy an agent skill that generates executive reports with live charts and KPIs.

2026-03-06
Agent skills are token optimized and highly efficient, but they break when you move to a new ecosystem. Platform-agnostic junctions fix that.

2026-03-04
Three platforms, three distinct approaches, one clear winner that captures actual market signal instead of speculative noise.

2026-02-28
For the 9th instalment of Cursor Meetup in Canada's Capital, I transformed the format from surface level AI exploration into automation mechanics.

2026-02-25
Terminal native AI agents now execute decisions, manipulate infrastructure, and manage entire systems without human intermediation.
Recent Events