Experience
Select a technology to see where I used it.
Full Stack Developer, Growth Engineering
Apertera- Build and maintain production software, internal tools and data applications in Python, JavaScript and PHP, with APIs, PostgreSQL and cloud services.
- Build internal applications in React, TypeScript, Node.js and PostgreSQL on Supabase and Vercel, with authentication and role-based workflows.
- Develop LLM-powered features and agentic automation that combine prompts, structured outputs, business rules and API orchestration.
- Prototype grounded AI that retrieves internal content and answers in context, using RAG, embeddings and semantic retrieval.
- Troubleshoot production across application, database and infrastructure: performance, caching, server configuration and data issues.
Full Stack Developer
Attributy Inc.- Developed scalable backend services for an advertising attribution and campaign-management platform in Node.js and TypeScript.
- Designed and maintained REST APIs and serverless services on AWS Lambda and API Gateway.
- Implemented data layers in PostgreSQL and DynamoDB for tracking, targeting and optimization.
- Built CI/CD pipelines with CircleCI and monitored production services.
Co-founder
Skano.Menu- Co-founded a QR-based digital menu platform: product direction, customer acquisition, analytics and features built from restaurants' needs.
Consulting Software Engineer
Albi Group- E-commerce in React, Node.js and WordPress, with product, payment and customer-account workflows, and Node.js automation for operations.
Consulting Software Engineer, Full Stack
Selmans Network- Web applications and e-commerce in Laravel, Django, JavaScript and Vue.js, with REST services, payment and shipping integrations and access control.
Software Developer
VentureUP- Responsive WordPress sites, custom themes, integrations and analytics tooling for startup projects, plus production maintenance.
Selected projects
RAG knowledge base and semantic search
LLM answers grounded in internal content, with their sources.
A retrieval-augmented knowledge system: embeddings and semantic retrieval surface the relevant source material, and the model answers from it. Structured metadata and multilingual search were part of the design.
LLMs, retrieval-augmented generation, embeddings, vector search
Second Read
Daily literature monitoring where every AI answer is checked twice.
Each morning it reads new PubMed papers on one topic. One model extracts the facts, plain code checks every fact is quoted word for word, and a second model from a different provider re-reads the paper. Only what passes both is accepted; the rest goes to a person with the reason.
Runs every morning.
Python, PubMed E-utilities, two LLMs from different providers, scheduled runs, Vercel
AI-assisted campaign analysis platform
Campaign data and scoring rules turned into short recommendations.
An internal platform with persistent application data and scoring logic. Structured LLM output combines business context and metrics into concise recommendations.
TypeScript, PostgreSQL on Supabase, LLM workflows, APIs, Vercel
Deep learning research: implicit bias of optimizers
How the choice of optimizer shapes what a model learns.
Controlled experiments comparing gradient descent with sharpness-aware optimization variants in linear and logistic settings, analysing optimizer behaviour and implicit bias.
Deep learning, gradient-based optimization, experiment design
Skills
- Languages
- Python, TypeScript, JavaScript, PHP, SQL
- AI and ML
- LLM integrations, retrieval-augmented generation, embeddings, semantic and vector search, agentic workflows, tool calling, structured outputs, prompt engineering
- ML research
- Deep learning, gradient-based optimization, experiment design, linear and logistic models
- Backend and cloud
- Node.js, REST APIs, webhooks, AWS Lambda, API Gateway, Docker, CI/CD, CircleCI, Vercel
- Data
- PostgreSQL, Supabase, DynamoDB, vector-enabled retrieval
- Frameworks
- Django, Laravel, React, Vue.js, WordPress
Education and languages
- 2015 to 2020
- BSc Computer Engineering, University of Prishtina, Faculty of Electrical and Computer Engineering
- Languages
- Albanian (native), English (professional), Italian and Spanish (working knowledge), German (learning)
Contact
The quickest way to reach me is by email.