About Amirhossein
Machine Learning Engineer with 3 years of hands-on experience designing and deploying production-grade AI systems across NLP, computer vision, generative AI, and agentic architectures. Proven track record building end-to-end intelligent systems — from advanced RAG pipelines and LLM fine-tuning to multi-agent frameworks with autonomous reasoning, tool use, and memory. Experienced both in-house and as a freelance AI lead, having independently led a team to deliver an agentic veterinary chatbot from concept to production. Proficient in Python, PyTorch, LangGraph, LangChain, Transformers, and CI/CD practices. Known for combining solid engineering discipline with a passion for innovation to deliver scalable, real-world AI solutions.
English
Native or bilingual
Italian
Conversational
Persian
Native or bilingual
Can work on-site
Turin (up to 50km)
Experience
- BV'TECHMachine Learning EngineerSeptember 2025 - Today (11 months)Milan, Metropolitan City of Milan, Italy• • Delivered Generative AI solutions for public sector and healthcare projects, covering document understanding, intelligent retrieval, and autonomous AI workflows.• • Designed and implemented agentic AI systems using LangGraph — with tool-calling, planning, and dynamic state management — automating multi-step document processing pipelines for government clients and reducing manual review effort.• • Built multi-agent orchestration workflows where specialized agents handled document routing, validation, compliance checking, and response generation, enabling end-to-end automated government document processing.• • Developed satellite image segmentation pipelines using transformer-based models (SegFormer, SAM) for land-use classification and infrastructure monitoring, supporting public-sector geospatial analysis projects.• • Developed and deployed document classification and analysis pipelines using LLMs and embedding models, enabling automated categorization, evaluation, and cross-comparison of government-submitted documents.• • Designed and improved RAG pipelines for hospital data, enhancing clinical information retrieval and response reliability with hybrid dense/sparse retrieval strategies.• • Fine-tuned language and embedding models on legal and healthcare domain terminology, measurably improving accuracy on specialized benchmarks.• • Enhanced the healthcare RAG system with a feedback loop and persistent memory management, enabling context retention and continuous refinement of model responses across sessions.• • Collaborated with clinicians, data engineers, and compliance teams to align GenAI system outputs with privacy, security, and regulatory requirements (GDPR, Italian healthcare standards). Technologies: LLMs, RAG, Agentic AI, LangGraph, Multi-Agent Systems, Embeddings, Model Fine-Tuning, Satellite Image Segmentation, SegFormer, SAM, Python, FastAPI, Vector Databases (FAISS/pgvector), PostgreSQL, Docker, CI/CD, Azure/GCP
- Pet24-7Freelance Lead AI EngineerJanuary 2025 - May 2026 (1 year and 4 months)London, UK• • Led a cross-functional AI engineering team to architect and ship a production-ready agentic veterinary chatbot, providing intelligent clinical decision support to veterinary professionals across the UK.• • Designed a multi-agent system (LangGraph) with specialized agents for symptom triage, drug-interaction lookup, treatment recommendations, and appointment coordination — enabling autonomous end-to-end veterinary consultation flows.• • Built an advanced hybrid RAG pipeline combining dense and sparse retrieval over a curated veterinary knowledge base, significantly reducing hallucination rates on clinical queries and improving response accuracy.• • Fine-tuned domain-adapted LLMs on veterinary medical literature, clinical guidelines, and SOAP note formats, improving diagnostic suggestion quality for both common and complex cases.• • Implemented persistent agent memory and multi-turn context management, enabling continuous clinical conversations with full session-level context retention across interactions.• • Established safety and evaluation frameworks — automated output testing, confidence thresholds, and human-in-the-loop review gates — to meet quality and safety standards for AI-assisted veterinary guidance.• • Integrated the system with practice management software APIs, enabling real-time patient record lookup and appointment booking directly within the chat interface. Technologies: LangGraph, Multi-Agent Systems, RAG, LLM Fine-Tuning, Agentic AI, Python, FastAPI, Vector Databases (pgvector/FAISS), PostgreSQL, Docker, AWS, CI/CD
- LogbotMachine Learning EngineerAugust 2024 - August 2025 (1 year)Padua, Province of Padua, Italy• • Developed a deep learning-based anomaly detection system for IoT sensors, identifying unusual patterns and reducing technician response time.• • Built a conversational AI interface (RAG-based) to simplify anomaly notifications and queries for technical staff.• • Fine-tuned BERT-based retriever models using contrastive learning, enhancing the relevance of responses by approximately 25%.• • Designed and deployed automated ML pipelines with CI/CD, ensuring reliable and timely model updates.• • Developed backend services and RESTful APIs to seamlessly integrate AI models into production workflows.• • Collaborated with multidisciplinary teams to integrate sensor data with visual analysis, improving overall diagnostic accuracy. Technologies: PyTorch, Deep Learning, Anomaly Detection, RAG, BERT, Contrastive Learning, Time-Series Analysis, ML Pipelines, Generative AI, Python, Docker, GitLab CI/CD, REST APIs
Recommendations
Be the first to recommend Amirhossein
Help this freelancer shine by sharing your experience working together.
These freelancer profiles also match your criteria
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Education
- Master of EngineeringUniversity of Padova2024Master of Engineering
- Bachelor in Computer EngineeringUniversity of Bonab2022Bachelor in Computer Engineering