ZEVQORA
An autonomous optimization engineer for AI applications: find expensive AI work, propose a cheaper execution path, verify it with evidence, and prepare the change for review.
Open project ↗Co-founder & CTO at ZEVQORA. I work across AI infrastructure, backend systems, defensive security, and developer tooling — turning rough technical ideas into systems that can be tested, explained, and shipped.
Most projects begin as a messy idea.
I care about architecture, APIs, reliability, security, evaluation, testing, and debugging — but mostly about whether a system can explain what it is doing and why.
The work is strongest when the clever part is invisible: fewer surprises, clearer boundaries, better evidence, and a product people can actually use.
ZEVQORA
ZEVQORA is an autonomous optimization engineer for AI applications. It finds expensive or unnecessary AI execution, proposes cheaper execution strategies, verifies the change through tests, replay, and quality evidence, then prepares reviewable code changes. Human review stays mandatory.
Visit ZEVQORA ↗A small archive of projects that show how I think across AI, backend, security, forensics, and interactive software.
An autonomous optimization engineer for AI applications: find expensive AI work, propose a cheaper execution path, verify it with evidence, and prepare the change for review.
Open project ↗A static URL risk analysis API that checks signals such as HTTPS, suspicious keywords, long URLs, IP-based domains, suspicious TLDs, shorteners, and brand-impersonation hints, then explains the risk clearly.
Open repository ↗A Python/Pygame survival runner focused on movement, collision detection, scoring, gameplay loops, and interactive system design.
Open repository ↗A NOtFound_404 team concept exploring obstacle detection, Bluetooth communication, mobile interaction, and Mongolian voice warnings for visually impaired users.
Open project ↗The principles are simple. Keeping them true in real systems is the hard part.
A faster wrong answer is still wrong. Measure the system first, then decide what should change.
Logs, boundaries, reports, and clear state transitions are part of the product — not cleanup work for later.
Defensive thinking works best when it shapes the design early, instead of being added after the system is already fragile.
Build the smallest version that proves the idea, test the weak parts, document what changed, and keep moving.
No fake finish line. Just a clearer direction with every project.
Learning to think through code, debugging, and small systems.
Understanding evidence, risk, and the discipline of authorized security work.
Moving from scripts into structured APIs, stored state, reports, and cleaner architecture.
Connecting the technical core to usable interfaces, documentation, deployment, and iteration.
Building the technical foundation for an AI optimization product with verification and human review at the center.
AI systems, backend engineering, defensive security, developer tooling, and ambitious technical collaborations.
khuslen.g784@gmail.com ↗