About

Hi, I’m Hari

I’m endlessly curious about how things work, from neural networks to neurons, from GPU kernels to economies.

What I Do

I build AI systems that bridge research and reality. Right now I’m a Senior Member of Technical Staff at Didero, where I architected the agent platform behind its procurement automation and the feedback loop that keeps it improving. Before that I worked on 3D reconstruction at Geopipe, and on multi-agent research systems at Andiron AI.

My research started with probabilistic models for decision-making. I invented Recurrent Sum-Product-Max Networks, published at ICAPS and IJCAI, and I’ve since worked across reinforcement learning, computer vision, LLMs and world models. What drives me most is the learning itself: the moment a complex idea suddenly clicks.

More on all of it on the Work page and in my CV.

What Fascinates Me

My curiosity goes well beyond code. I’m drawn to:

  • The nature of reality. From the Big Bang to quantum mechanics, from general relativity to the emergence of biological complexity. How did we get here?
  • Mathematics and computation. Gödel’s incompleteness theorems, Turing machines, lambda calculus, Bayesian networks and scaling laws. The foundations that make intelligence possible.
  • Consciousness and philosophy. How do neurons create experience? What did ancient philosophers mean by Brahman? I explore everything from Advaita Vedanta to the hard problem of consciousness.
  • Complex systems. Economics, politics, neuroscience: anywhere patterns emerge from chaos. How do markets organize themselves? Why do civilizations rise and fall?
  • Movement and stillness. I dance to think with my body and express what words can’t. I practice Vipassana meditation, and a 10-day silent retreat rewired how I see reality.

My approach blends contradictions. I meditate yet stay optimistically anxious about humanity’s future. I build rigorous mathematical models while believing consciousness may transcend computation. I obsess over technical details but know that in the end, human taste and intuition are what matter most.

Current Obsessions

  • Agent memory that knows how sure it is, and learns which sources to trust
  • Economics from the ground up, and what AI will do to work and value
  • World models, and when they actually help agents plan
  • How consciousness relates to computation

Let’s Connect

I love conversations that challenge assumptions. Whether it’s agent architectures, economic systems or the nature of consciousness, if you’re curious about it, I probably am too.

Email · LinkedIn · GitHub