Cristian Cirje

Software and AI engineer

Cristian Cirje

About

Computer Engineering and Artificial Intelligence undergraduate at Roma Tre University, currently CTO of GeniusHub — an AI-assisted study platform for university students. Hands-on experience building distributed, multi-tier systems end to end, from low-level systems programming in C to backend services, relational databases, and production web applications used by real users. Most drawn to applied AI: RAG pipelines, LLM-powered products, and the infrastructure that runs them.

Projects

  • Studio AI

    2025 — Present
    AI agent embedded in FL Studio for organizing music-production projects through natural language. A multi-tier distributed system: a native Rust plugin inside the DAW, a Python IPC bridge, and a real-time FastAPI relay service that talks to a Next.js dashboard over WebSockets.
    Next.jsReactFastAPIRustPythonWebSocketsGoogle Gemini

  • picoforge

    In progress
    Single-model LLM inference engine for Apple Silicon, written from scratch in C and Metal with no ML frameworks. Every stage is validated against a NumPy reference and a C CPU baseline before the Metal GPU kernels are written — catching numerical drift early and keeping memory bandwidth and compute complexity in view.
    CMetal 4Apple SiliconPythonNumPy

  • RelyRAG

    2025
    Retrieval-augmented generation platform for querying uploaded documents. An async ingestion pipeline — background jobs, chunking, and embeddings — feeds a vector index on PostgreSQL, paired with a streaming retrieval-and-generate query flow with query rewriting.
    Next.jsTypeScriptPostgreSQLpgvectorGoogle GeminiInngest

  • Zenthes

    2024 — 2025
    Full-stack SaaS platform with subscription billing, license management, and end-to-end type-safe APIs. Stripe-powered payments and subscriptions with webhook-driven license activation and lifecycle management.
    Next.jsTypeScriptPostgreSQLPrismatRPCStripe

Technical skills

Programming
TypeScriptJavaScriptPythonCJavaSQLHTMLCSS
Web & Backend
Next.jsReactNode.jsFastAPItRPCRESTWebSocketsTanStack QueryDrizzle ORMPrismaTailwind CSS
AI & Machine Learning
LLM application developmentRAG pipelinesEmbeddings & rerankingPrompt engineeringVector databasesVercel AI SDKPyTorchTensorFlow / Keras
Data & Infrastructure
PostgreSQLSupabaseMongoDBRedisAWSVercelGit
CS Fundamentals
Data structures & algorithmsComputational complexityObject-oriented designSystem design

Work experience

  • GeniusHub logo

    CTOGeniusHub

    Jun 2026 — Present

    Technical lead of GeniusHub (formerly GeniusEdu), an AI-assisted study platform for university students — owning architecture, data model, and delivery priorities end to end. Designed and built the RAG pipeline: multi-format ingestion, contextual chunking, embeddings, and retrieval with reranking, with answers backed by verifiable citations. Shipped features from schema to UI, including chat over study materials, AI-generated summaries and concept maps, auto-graded mock exams, and subscription billing.

  • Full-Stack EngineerFreelance

    2024 — Present

    Designing and shipping production web applications for small businesses and early-stage founders, end to end from data model to deployment — including romanadvisor.com, built on Next.js and Supabase with Redis-based rate limiting, a moderated review system, and scheduled jobs. Fast iteration cycles, clean handoff, and production-grade defaults across multiple client engagements.

  • AI Kosmo logo

    Full-Stack DeveloperAI Kosmo

    Sep 2025 — Oct 2025

    Shipped front-end features for the Kosmo Cloud dashboard, sold to hotel clients. Contributed to the codebase of a multi-channel LLM product spanning chat, voice, and in-room guest channels.

Education

  • Università degli Studi Roma Tre logo

    BSc, Computer Engineering & AIUniversità degli Studi Roma Tre

    Sep 2025 — Jul 2028

    Three-year program covering software development, algorithms and data structures, databases, computer networks, operating systems, and the foundations of artificial intelligence and machine learning. Strong mathematical and physics core — calculus, linear algebra, and probability.

  • Technical High School Diploma, Computer ScienceITC Vincenzo Arangio Ruiz

    Sep 2020 — Jul 2025

    Five-year technical program in computer science — programming (Java, C/C++, PHP, JavaScript), web development, relational and NoSQL databases, and Cisco networking.