08.09.2026 aktualisiert
100 % verfügbar
Senior AI Engineer | LLM, RAG, Agentic Systems | Enterprise Modernization · Remote
Prishtina, Kosovo
Nur Remote
BSc Computer Science (Univ. of Prishtina); MSc Data Science studies (Universum College)Über mich
30+ years in enterprise software; since 2024 building production LLM, RAG and agentic systems with governed execution (RBAC, approvals, audit trails). Public-finance, tax and legacy modernization for Deloitte, World Bank, GOPA. Remote from CET, English, B2B via Inovativi SH.P.K.
Skills
JavaData MigrationEnterprise Architecture FrameworkPythonPostgreSQLMachine LearningOracle DatabasesSoftware ArchitectureSolution ArchitectureTypeScriptEnterprise Application IntegrationChatbotsLarge Language ModelsPrompt EngineeringGenerative AI
LLM Applications & RAG
Expertise in building production LLM applications using OpenAI and Anthropic APIs, open-weight and local models including Llama, Qwen, and Mistral via Ollama and LM Studio. Proficiency in embeddings, Qdrant vector search, retrieval design, structured outputs, prompt engineering, tool calling, MCP, document AI, and provider abstraction.
Agentic & Enterprise Systems
Design and implementation of agentic systems using PLAN to EXECUTE to COMPOSE orchestration, permission-filtered tools, sub-workflow design, RBAC, approvals, audit trails, idempotency, validation, shadow systems, reconciliation, data migration, and legacy integration.
Evaluation & Reliability
Development of custom LLM evaluation harnesses, metrics and benchmark cases, repeated runs, baseline and guarded A/B comparisons, prompt regression, groundedness, abstention, quality gates, deterministic mocks, automated testing, and observability.
Applied ML & NLP
Applied machine learning and NLP foundation using Python, pandas, NumPy, scikit-learn, and PyTorch. Experience in data preparation, feature engineering, time-series analysis, regression, classification, clustering, model evaluation, NLP, text processing, and embeddings.
Backend & API Development
Backend development with FastAPI, TypeScript, Next.js, NestJS, Java, Spring Boot, REST and OpenAPI, WebSockets, queues, PostgreSQL, Oracle, SQL Server, Redis, Qdrant, MinIO, Docker, Vercel deployments, and Linux.
AI-Native Delivery
Delivery using Claude Code, Codex, workflow discovery, rapid vertical slices, direct client implementation, feedback loops, technical documentation, training, and operational handover.
Enterprise & Solution Architecture
Enterprise and solution architecture covering event-driven and modular systems, revenue administration, public finance, HR, legal information, real estate, CRM, DACH sales and marketing operations, e-services, and SME workflows.
Digital Government & Public Finance Systems
Delivery of tax-administration and digital-government systems including SIGTAS, e-filing, e-payment, reconciliation, vehicle and driver services, HR, data migration, and enterprise integration across multiple international engagements.
Expertise in building production LLM applications using OpenAI and Anthropic APIs, open-weight and local models including Llama, Qwen, and Mistral via Ollama and LM Studio. Proficiency in embeddings, Qdrant vector search, retrieval design, structured outputs, prompt engineering, tool calling, MCP, document AI, and provider abstraction.
Agentic & Enterprise Systems
Design and implementation of agentic systems using PLAN to EXECUTE to COMPOSE orchestration, permission-filtered tools, sub-workflow design, RBAC, approvals, audit trails, idempotency, validation, shadow systems, reconciliation, data migration, and legacy integration.
Evaluation & Reliability
Development of custom LLM evaluation harnesses, metrics and benchmark cases, repeated runs, baseline and guarded A/B comparisons, prompt regression, groundedness, abstention, quality gates, deterministic mocks, automated testing, and observability.
Applied ML & NLP
Applied machine learning and NLP foundation using Python, pandas, NumPy, scikit-learn, and PyTorch. Experience in data preparation, feature engineering, time-series analysis, regression, classification, clustering, model evaluation, NLP, text processing, and embeddings.
Backend & API Development
Backend development with FastAPI, TypeScript, Next.js, NestJS, Java, Spring Boot, REST and OpenAPI, WebSockets, queues, PostgreSQL, Oracle, SQL Server, Redis, Qdrant, MinIO, Docker, Vercel deployments, and Linux.
AI-Native Delivery
Delivery using Claude Code, Codex, workflow discovery, rapid vertical slices, direct client implementation, feedback loops, technical documentation, training, and operational handover.
Enterprise & Solution Architecture
Enterprise and solution architecture covering event-driven and modular systems, revenue administration, public finance, HR, legal information, real estate, CRM, DACH sales and marketing operations, e-services, and SME workflows.
Digital Government & Public Finance Systems
Delivery of tax-administration and digital-government systems including SIGTAS, e-filing, e-payment, reconciliation, vehicle and driver services, HR, data migration, and enterprise integration across multiple international engagements.
Erfahrung
Berufserfahrungmehr als 10 Jahre
Freelancer seit2007
Sprachen
EnglishverhandlungssicherSerbianverhandlungssicherAlbanianMuttersprache
Projekthistorie
Building production LLM applications, RAG pipelines, and agentic workflows using OpenAI, Anthropic, and open-weight/local models. End-to-end delivery from prototype through Python/FastAPI and TypeScript integration, automated testing, deployment, monitoring, user validation, and technical handover. Designing controlled execution layers connecting models to databases, APIs, CRM/admin tools, documents, messaging channels, and legacy systems. Building repeatable LLM evaluation suites and modernizing public-finance and HR platforms.
Delivered full-stack and systems engineering across web applications, backend services, integrations, data flows, deployment, troubleshooting, and production support, collaborating in a two-person engineering team with shared implementation, code-review, and delivery responsibility.
Architected systems for German-market sales operations, CRM, workflow coordination, reporting, analytics, and integration.