Looking for a job

Chaima Gharbi

@chaimagharbi

AI Engineer at Vamisec who built a multi-agent compliance analysis engine, cutting first-pass analysis from about two weeks to under three hours.

Germany

Message

At Vamisec, I built a framework-agnostic multi-agent LLM engine that assesses policies against standards including ISO 27001/27701, NIST, and DORA. It returns control-level verdicts with confidence, rationale, and quoted evidence.

I designed its Reasoner–Critic–Router architecture and hybrid RAG pipeline, with safeguards to ground answers in evidence. The system cut first-pass analysis from about two weeks to under three hours, with about 95% agreement with expert labels.

At Mercedes-Benz, I fine-tuned transformer models for semantic search in defect tickets, improving retrieval accuracy by about 25%. I also integrated MCP tools into a GPT-based LangGraph agent for ticket-related queries.

At Hexastack, I built CI/CD pipelines for experiments and Dockerized NLP model servers, and implemented DVC pipelines for model and dataset versioning. My projects include an MCP-based DevOps agent, credit card fraud detection, and MRI bone detection using Meta’s Segment Anything Model.

Experience

Work history, roles, and key accomplishments

AI Engineer Intern

Vamisec

Mar 2026 - Aug 2026 (5 months)

Built a framework-agnostic multi-agent LLM engine for automated compliance gap analysis, and designed a Reasoner–Critic–Router architecture over a hybrid RAG pipeline. Engineered grounding and anti-hallucination safeguards, and built the product's audit interface in TypeScript/React.

LLM Multi Agent Systems RAG Pgvector PostgreSQL Cross Encoder Re Ranking Langfuse TypeScript React ISO 27001 NIST DORA OSCAL

NLP Intern

Mercedes-Benz

Jul 2025 - Sep 2025 (2 months)

Fine-tuned transformer models for semantic search in defect tickets on Databricks, improving retrieval accuracy by ~25%. Integrated Model Context Protocol (MCP) tools into a GPT-based LangGraph agent with Langfuse tracing.

Transformers Fine Tuning Semantic Search Databricks MCP LangGraph Langfuse GPT

MLOps Intern

Hexastack

Jul 2024 - Aug 2024 (1 month)

Built CI/CD pipelines to automate experiments and deploy Dockerized NLP model servers to Hugging Face. Implemented DVC pipelines for model and dataset versioning.

CI CD Docker Hugging Face DVC MLOps NLP

Education

Degrees, certifications, and relevant coursework

National Institute of Applied Science and Technology (INSAT)

Engineer's Degree, Software Engineering

2021 -

Pursuing an Engineer's Degree in Software Engineering, expected to complete in September 2026.

Coursera

Specialization, Machine Learning and NLP

Completed Machine Learning and NLP Specializations covering supervised and unsupervised learning, recommenders, reinforcement learning, classification, probabilistic models, sequence and attention models.