Aditya Naganath brings an unusually deep technical background to venture investing, having worked as a technical lead at Palantir, software engineer at Twitter and Nextdoor, and product manager at Google, giving him hands-on experience across the full engineering and product stack. He focuses on AI infrastructure with research-level expertise in areas like GPU utilization, FlashAttention, and LLM training workloads, and has developed a clear thesis distinguishing true infrastructure layers from higher-level tooling, backing companies like Together AI based on his conviction that open-source models will be good enough for enterprises.

• Deep interest in the strategic evolution of AI startups, specifically emphasizing that successful applications will capture both software and labor budgets by reducing headcount while supercharging remaining workers. A key belief is that founders must use counter positioning against incumbents and transition from basic prompt engineering to advanced context engineering to better retrieve domain specific information.
• A focus on the practical deployment and user experience of agentic AI. For example, noting that AI in accounting requires deterministic execution and human review because plausible answers are not enough for such unforgiving work. Another major observation is that broad adoption of agents is currently limited by infrastructure rather than model capabilities, alongside a need for basic product improvements like in line editing to avoid constant output regenerations.
• Appreciation for the fundamental qualities of exceptional founding teams and the realities of early stage venture capital. The core investing philosophy is described as finding a wonderful team at a fair price, with high regard for founders who can combine massive big picture thinking with strategic on the ground execution. There is also an understanding that ambitious ideas take extensive time to mature, referencing how long it took for Einstein to refine general relativity before it became the gold standard.
• Enthusiasm for enterprise partnerships and innovative ways software startups can scale. An interesting reference point is Ramp, which is praised for its approach to design partnerships where they test and build with early stage AI tools to drive their own internal productivity. This ties into a broader goal of backing products that democratize access to business insights across entire organizations, a passion tracing back to earlier career experiences at Palantir.