Carolina Leguizamon
Applied AI Engineer with a background in electronics and firmware. That background shapes how I build AI systems — state machines, circuit breakers, watchdog timers, checkpointing. The patterns that keep embedded systems running are the same ones that keep AI systems honest in production.
I focus on production RAG, document intelligence pipelines, and agentic workflows — the systems that have to keep working after the demo.
- AI Platform EngineerWalma AIAI engineering on Walma's distributed platform (Noda + Ocle codebases). Rebuilt the municipal Invoice Reviewer (Fakturagranskare) from 1/11 to 11/11 reliable outputs, established the Ocle/Noda architectural split, migrated embeddings from ada-002 to text-embedding-3-small, and consolidated infrastructure onto a shared Azure AI Foundry hub.2026
- Content StrategyWalma AIBuilt Walma's content presence from scratch: 29 thought-leadership articles, LinkedIn strategy across 14 topic areas, and an analytics-led repositioning. +5,766% impressions and search appearances from ~39/week to 800–1,000/week over 46 days — fully organic.2026
- Applied AI EngineerVermiculus Financial TechnologyEnterprise RAG built from scratch and deployed on-prem — hybrid retrieval (BM25 + pgvector), RRF, cross-encoder reranking, Instructor-XL embeddings, Qwen 32B locally on NVIDIA GPU. −67% time-to-information.2025 — 2026
- AI Consultant (Independent)Di LucaroLLM visibility strategies, retrieval-aware content architectures, analytics dashboards, and end-to-end web builds for founder-led companies and independent brands.2025 — Present
- Customer Data Developer (YH)IHM Business School2026
- Python for AI · ReactIT-Högskolan2024
- Full-Stack Dev · ML & Predictive AnalyticsCode Institute2024