Ilyros

AI workflows & software engineering
By Jasin Marku · New York

Discuss your workflow inquiries@ilyros.com

Better workflows.
Practical engineering.

I’m Jasin Marku, the independent engineer behind Ilyros, based in New York City. My background runs from systems administration to building internal tools and production applications. That experience shapes how I approach AI workflows and automation: understand the work, connect the right systems, and take responsibility for getting the result into use.

Jasin Marku

Software Engineer · New York City

LinkedIn

Technical skills

The tools and practices behind my application, data, and infrastructure work.

Languages
Python, TypeScript, JavaScript, SQL, Bash
Full-stack development
React, FastAPI, Node.js, REST APIs, SQLAlchemy, Pydantic
Data & integrations
PostgreSQL, DuckDB, Amazon Redshift, Salesforce, API integrations
Cloud & DevOps
AWS (ECS, Fargate, EC2, Lambda, S3, CDK), Docker, GitHub Actions, CI/CD
Security & reliability
OAuth, OIDC, IAM, role-based access control, CloudWatch, production monitoring
Testing
pytest, Vitest, Playwright, Locust, integration testing, performance testing

Professional experience

My roles and selected contributions, most recent first. The enterprise and internal operations work here is professional employment experience, separate from projects delivered through Ilyros.

Software Engineer, Forward Deployed AI

I build full-stack applications and workflow systems in an enterprise environment, working directly with stakeholders from requirements through deployment, production support, and release.

Operational approvals platform

I engineered and launched a React, FastAPI, and PostgreSQL platform connecting eligibility rules, routing, and administrative approvals. Integrations with an enterprise warehouse, Salesforce, and email delivery close the loop between a request, its decision, and the next action.

The workflow removed 700+ stale requests from staff queues through automated notifications, approval workflows, and case reconciliation.

I owned deployment from pre-production through an initial pilot, including role-based access, immutable audit histories, automated testing, backups, and production monitoring.

Enterprise analytics & business applications

I delivered five full-stack applications to production in a single month. The analytics and business applications were designed to support 2,700+ internal users. I architected and deployed Python, FastAPI, React, and TypeScript applications on AWS, integrating warehouse data, external APIs, and single sign-on.

The data work also includes scheduled snapshots and precomputed datasets, with parity checks before data reaches the interface. This moves repeated preparation ahead of dashboard requests.

  • I reduced critical page-load latency from 125 seconds to 1.1 seconds (p95) through caching and asynchronous processing.
  • I validated up to 4,000 concurrent users with zero failures in load testing.
  • I supported 100+ releases and 6,000+ automated tests across multiple applications.

These results describe specific professional work and test conditions; the single-month delivery period is a past example.

Deployment, access & reliability

My work includes CI/CD pipelines, security controls, monitoring, incident resolution, and stakeholder-reviewed releases. Ownership varies by system: for one earlier internal application, I owned infrastructure, deployment, runbooks, and handoff rather than its full business logic.

Software Engineer

I built internal React and FastAPI dashboards and Python data pipelines to bring project tracking, costs, and financial reporting into PostgreSQL, replacing manual spreadsheet workflows.

I automated SQL reporting for project managers and finance teams, and led a company website rebuild from architecture through production deployment on Cloudflare Pages.

Systems Administrator

I administered Microsoft 365 and Azure Active Directory for 20+ users, managing identity and role-based access. I completed a company-wide multi-factor authentication rollout with zero account lockouts. That work gave me a practical grounding in access, reliability, and the people depending on the system.

Independent projects through Ilyros

Ilyros is where I bring that engineering experience to independent products and client work. You work directly with me on scope, design, implementation, and delivery.

VirtuTrade · Native iOS

I designed, built, and published VirtuTrade as its sole engineer. The paper-trading app brings live market data for 250 assets together with persistent portfolios, watchlists, and performance tracking. Trading uses virtual funds.

Built with Swift, SwiftUI, Combine, and Core Data, it includes trade validation, cash reconciliation, transaction history, and networking with retries, TTL caching, and stale-data fallbacks.

Explore VirtuTrade

I also build responsive websites, applications, dashboards, and internal tools. View selected websites or request additional relevant work.

Education & certification

B.A. in Computer Science
Cum Laude
AWS Certified Cloud Practitioner
Issued February 2026

AI in the workflow. AI in the build.

AI with a review path.

My experience with AI-assisted workflows includes structured model outputs, evaluation, and auditable human overrides. I define the task, the expected output, and where someone needs to review or take over. The choice between AI, a simpler integration, and conventional software follows the problem.

In my development process, I use AI tools and coding agents with scoped tasks, repository instructions, and automated checks to implement and iterate quickly. I remain responsible for architecture, integrations, debugging, and verifying what reaches production. Documentation and handover are part of finishing the work.

What’s slowing the work down?

A repetitive task, disconnected tools, or a product you need built.
Tell me what’s happening. I’ll help define the next step.