Marcello De Bernardi
| Marcello De Bernardi | |
| Occupation | Co-founder and CTO of Plexe AI |
|---|---|
| Known for | Plexe AI, loss-landscapes open-source library |
Marcello De Bernardi operates as a technology executive in London. He co-founded Plexe AI with Vaibhav Dubey and holds the chief technology officer (CTO) position. The firm builds agents. Operators type natural language prompts directly on the platform interface to create predictive machine learning models.[1]
Career
Before accepting executive engineering positions, De Bernardi built a technical foundation writing code for the open-source machine learning community. He programmed "loss-landscapes," an open-source tool, in 2019. The software approximates and visualizes the loss landscapes of advanced neural networks. The repository gained popularity among machine learning researchers. Academic computer science literature cites the project in publications such as CoPhy-PGNN: Learning Physics-guided Neural Networks, available from the Association for Computing Machinery (ACM) Digital Library.[2]
Prior to establishing Plexe, De Bernardi spent approximately nine years at Expedia Group. His tenure featured multiple roles of increasing seniority like ML Scientist III, ML Engineer III, and Manager of ML Engineering. He managed production-grade machine learning systems at scale, a responsibility that explicitly involved overseeing intricate data pipelines and directing engineering teams to deploy predictive algorithmic models for enterprise environments. Previous positions also included software work at Amazon Web Services (AWS). That job required developing data science and software engineering tools focused on stable cloud infrastructure for machine learning.[3]
Plexe AI accelerates the creation of custom machine learning models. Setup moves fast. Operators connect data repositories before outlining desired system specifications using plain language. Autonomous agents handle the necessary technical steps autonomously, executing backend workflows that range from initial data quality checks and pattern recognition to final model training and deployment. Packaged software components make machine learning accessible for enterprises lacking deep data science expertise. Essential technical elements include data dashboards, API endpoints, batch jobs, file upload access, and database connectors.[4]
Plexe participated in the inaugural Spring 2025 (X25) batch at Y Combinator. After Demo Day, Business Insider highlighted the firm as a highly anticipated AI agent startup from the cohort.[5][6] The company secured a $2.1 million seed funding round in August 2025. The business plans expansion efforts targeting the United States market. Future roadmaps prioritize the retail sector plus banking, financial services, and insurance (BFSI) clients to attract an upcoming Series A round.[7]
Logistics and manufacturing enterprises use company architecture to build autonomous supply chain agents, deploying applied artificial intelligence to fully automate procurement, advanced inventory management, and supplier coordination.
References
- ↑ "AI startup Plexe debuts: Natural language meets autonomous agents for next-gen machine learning". 'The American Bazaar}'. Retrieved 2026-03-19.
- ↑ "CoPhy-PGNN: Learning Physics-guided Neural Networks", ACM Digital Library, 2022.
- ↑ "Marcello De Bernardi, Co-Founder CTO at Plexe AI". 'Prog.AI}'. Retrieved 2026-03-19.
- ↑ "Plexe AI". 'Plexe}'. Retrieved 2026-03-19.
- ↑ "These are 10 of the most exciting AI agent startups to come out of Y Combinator's first-ever Spring batch". 'Business Insider}'. Retrieved 2026-03-19.
- ↑ "Y Combinator X25 Batch Companies", Extruct AI, Retrieved March 2026.
- ↑ "Plexe secures $2.1m seed to expand AI agents for enterprises, eyes Series A next". 'The Telegraph India}'. Retrieved 2026-03-19.