# Ajit Koti

> at the intersection of distributed systems, agent infrastructure, and early-stage investing

Engineer · Investor · Protocol Designer. Based in San Jose, California.

I build distributed systems for a living — currently the Identity Graph at Netflix.
I design open protocols for how autonomous agents coordinate and trust each other.
I back early-stage founders building the infrastructure layer underneath AI.

## Open Source

- [Multi-Agent Coordination Protocol](https://www.multiagentcoordinationprotocol.io/): An open specification for how autonomous agents negotiate task ownership, hand off work, and resolve conflicting goals without a central coordinator.
- [Agent Identity & Trust Protocol](https://www.agentidentitytrustprotocol.io/): A specification for verifiable agent identity and trust attestations, so agents can establish who they're talking to and how much to trust them.
- [Agent Context Distribution Protocol](https://www.agentcontextdistributionprotocol.io/): A specification for sharing and synchronizing context across agents and sessions without re-deriving it from scratch each time.

## Research

- [Parkinson's Variant Explorer](https://github.com/geneticfold/neuro-fold): Explores how Parkinson's-associated mutations in SNCA, LRRK2, PINK1 and PARK7 alter protein structure. Compares wild-type and mutant folds, quantifies the deviation by RMSD, renders both in an interactive py3Dmol viewer, and pairs each variant with an LLM agent that summarizes the relevant PubMed literature. Streamlit app, containerized.
- [ADPKD Variant Explorer](https://github.com/geneticfold/nephro-fold): Examines pathogenic PKD1 and PKD2 variants behind autosomal dominant polycystic kidney disease against AlphaFold-predicted structures. Pulls variant data from ClinVar and the PKD Mutation Database and renders the mutations in an interactive 3D viewer.

## Investing

- [TBD VC](https://www.tbd.vc/): Early-stage fund backing Israeli deep-tech founders at pre-seed and seed, across enterprise AI, cloud infrastructure, and cybersecurity. Venture partner alongside a network drawn from Netflix, GitHub, and Epic Games.
- [Sheva](https://sheva.vc/): Pre-seed and seed fund investing in Israeli startups across SaaS, cyber, fintech, and B2C, led by Omri Casspi and David Citron.
- [Anansi Capital](https://www.anansi.capital/): Solo-GP fund backing AI-enabled rollups of SMB and services businesses at pre-seed and seed, run by Vinay Iyengar.
- [Hustle Fund — Angel Squad](https://www.hustlefund.vc/squad): Angel investing alongside Hustle Fund's deal flow, co-investing in its pre-seed and seed portfolio.

## Theses

- [AI — Agents are a distributed systems problem](https://www.ajitkoti.com/theses): The hard part of multi-agent systems was never the intelligence. It is everything around the intelligence: how independent processes discover each other, agree on who owns a piece of work, prove they are who they claim to be, and share enough context to act coherently. We have spent two decades solving versions of these problems for services. We are now solving them again, badly, for agents.
- [Investing — Backing the layer underneath the layer](https://www.ajitkoti.com/theses): I invest in the infrastructure that AI applications will depend on and mostly never think about — the coordination, identity, data, and reliability plumbing. I do it as a venture partner at TBD VC and Sheva, as an LP in Anansi Capital, and as an angel through Hustle Fund's Angel Squad. Small checks, early, in the part of the stack I actually understand.

## Talks

- [Pattern Matching @ Scale Using Finite State Machines](https://www.youtube.com/watch?v=WykSdgtLDD0): Strange Loop, 2019
- [A Tale of Stateful Stream to Stream Processing](https://www.youtube.com/watch?v=pEH0kx6jK0M): Flink Forward
- [A Tale of Stateful Stream to Stream Processing](https://www.youtube.com/watch?v=ucwwTo62o9w): Reactive Summit
- [Finding Needles in Big Data Haystacks](https://www.youtube.com/watch?v=qyzVWq1aSUg): Scale By The Bay

## Writing

- [Built a Real-Time Distributed Graph: Part 2](https://netflixtechblog.medium.com/how-and-why-netflix-built-a-real-time-distributed-graph-part-2-building-a-scalable-storage-layer-ff4a8dbd3d1f): Building a scalable storage layer.
- [Built a Real-Time Distributed Graph: Part 3](https://netflixtechblog.medium.com/how-and-why-netflix-built-a-real-time-distributed-graph-part-3-querying-the-graph-with-grpc-0f3468349607): Querying the graph with gRPC.

## Elsewhere

- [Twitter](https://x.com/ajitkoti)
- [LinkedIn](https://www.linkedin.com/in/ajitkoti)
- [GitHub](https://github.com/ajitkoti)
- [Medium](https://medium.com/@ajitkoti)