I'm a Full-Stack AI Engineer with 3+ years of production experience building enterprise-scale AI systems. As the sole AI engineer at SIPA SpA (Zoppas Industries Group), I designed and productionised AI into ECHO Platform, an enterprise AI assistant ecosystem serving 3,000+ internal staff and customers globally, that gives after sales services using ERP, CRM, PLM, ticketing, and IoT data streams through three specialised AI agents.

My work spans the full lifecycle: problem framing → system architecture → retrieval design → agent orchestration → evaluation → production deployment → LLMOps observability. I care about systems that are measurable, explainable, and maintainable at scale.

Some projects below were built at SIPA SpA. Proprietary code lives on the company's internal GitLab.

Multi-Agent multi-step real-time query system connecting to 25+ live REST APIs (machines, orders, production data, alarms) via hierarchical intent classification → sub-agent identification parallell / sequential, delegate / handoff → parallel function calling → parameter extraction → parallel sub-agent handoffed LLMs answer synthesis to user parallely (or) return to orchestrator for next step plan -> Final answer sysnthesis. Integrated into the platform backend via WebSockets + JavaScript front-end, with RBAC data security for 3,000+ users. Includes Langfuse observability, real-time user feedback loops, and continuous RAG evaluation.

PydanticAI Django WebSockets Langfuse OpenAI PostgreSQL REST APIs RBAC

Built a conversational fleet intelligence system that predicts truck component failures, explains risk with SHAP, and supports maintenance decisions through an agentic chat interface backed by FastAPI services. The project combines PydanticAI, XGBoost, Langfuse, and a Scania-inspired fleet simulator, using real predictive-maintenance patterns from the Component X dataset.

PydanticAI FastAPI XGBoost SHAP Langfuse Predictive Maintenance Agentic Systems Simulator

Advanced hybrid RAG pipeline for industrial PDF manuals (maintenance, installation, operational). Parses hierarchical document structures via Table of Contents, stores chapter/sub-chapter hierarchies in PostgreSQL with dynamic linking to pgvector embeddings (OpenAI text-embedding-3-large, dim=256) in AWS S3. Implements FTS + vector search fused with Reciprocal Rank Fusion (RRF), LLM re-ranking, extractive summarisation with citations, and multi-turn query rewriting.

RAG pgvector PostgreSQL AWS S3 Hybrid Search RRF OpenAI Embeddings Langfuse Django

Data pipeline for historical service ticket retrieval: PII removal via a domain-specific technical glossary, structured metadata extraction (alarms, machine models, issue types), and embedding-based indexing. Dual-mode retrieval: deterministic query mode (structured ticket search by parameters) + RAG semantic mode, with cited ticket references for traceability.

RAG NLP Django PII Removal Metadata Extraction Semantic Search PostgreSQL LangChain Langfuse Docker

Developed a forecasting solution predicting alarm events and production cycle timing across 4,200 industrial machines at once, enabling proactive maintenance scheduling. Applied XGBoost + Transformer-based architectures (LSTM, ARIMA) on large-scale industrial sensor data.

XGBoost LSTM ARIMA Transformers Scikit-learn MLflow Time Series PySpark

▶ Watch the demo video

▶ Watch the demo video

▶ Watch the demo video

Architecture

▶ Watch the demo video

Architecture

Architecture

Screenshots

Generative AI & LLMs

Agentic & LLMOps

Backend & APIs

ML & Data Science

Databases & Storage

• 🏢 Full-Stack AI Engineer @ SIPA SpA (Zoppas Industries Group) - Italy
• 🎯 Open to senior AI/ML Engineer roles in Stockholm, Sweden or any other location in Sweden
• 🌐 Languages: English · Italian · Hindi · Telugu