Subho Majumdar, PhD
Vijil is on a mission to help enterprises build and operate intelligent agents that people can trust. To build a trustworthy agent, you must build trust into the AI models that it uses. For that, you must be able to inspect, modify, and control the models. This is easier to do when the models are open-source. Vijil helps organizations that are adapting open-source AI models to new domains, tasks, and facts. For AI developers customizing open-source large language models, Vijil provides tools to harden model
Enterprises cannot deploy generative AI in production today because they cannot trust custom LLMs to behave reliably in the real world. LLMs are prone to errors, open to threats, easy to breach, and slow to recover. Even if they were originally designed for honest and helpful use, they can be misused to produce hallucinations, toxic content, inexplicable outputs, and unfair outcomes.
LLMs are vulnerable to an unbounded set of attacks including prompt injection, model theft, data theft, model evasion, and data poisoning.
Subho Majumdar, PhD