pro
Vlad Ioan
AI Infrastructure Engineer | On-Premise LLM & RAG
New to Contra
Vlad is building their profile!
Built two production AI agents via Dify tool-calling: nmap network
scanner (natural language → structured scan results) and Active
Directory lookup (LDAP/NTLM auth, user/group queries). LLM
orchestration with Granite-4.0-Tiny as tool-calling model.
1
8
Built production RAG system using Dify 1.13.x, Qdrant vector
database, BGE-M3 embeddings and bge-reranker-v2-m3 reranker.
Chunk size 800/overlap 200. Integrated with internal document
repositories. Achieves reliable retrieval on air-gapped
infrastructure with no cloud dependency.
1
12
Deployed a multi-model CPU inference cluster on 2× HP ProLiant
DL360 Gen9 servers (376GB RAM total). Running 5 concurrent LLM
endpoints: Qwen3-Coder-30B, Qwen3-Next-80B, GLM-4.7-Flash,
Granite-4.0-Tiny — all quantized (GGUF/Q4_K_M). Optimized with
ik_llama + MKL for +62% throughput. Ollama-compatible API,
Open WebUI frontend, Opik observability.
1
17