About
Hey there. I am Kumar Shivendu. 👨💻 Here's a PDF version of my Resume
I ❤️ search (retrieval and LLMs), distributed databases, observability, and mining data. I'm particularly obsessed with retrieval as a problem, and I've been building and tinkering with search systems for a while:
- Search over 420M+ OSS repos at Software Heritage during college (GSoC).
- Reverse-engineered the architecture and cost of Exa using napkin math
- I'm one of the early engineers of Qdrant and I built distributed systems that power search at billion scale for companies like Canva, Perplexity, Meesho, Hubspot, etc., and run on almost a million machines worldwide.
I'm actively researching a new class of embeddings and ways to make agentic search systems faster and learn continually.
🔧 Some of my cool side projects:
- smoldb: A distributed database in Rust to improve programming taste, intuition, and CS fundamentals.
- Monitoring my life with Grafana, Prometheus, and InfluxDB: Health, time, and productivity data from my devices in Grafana dashboards.
- I Reverse-Engineered Exa.ai Infrastructure Cost with Napkin Math: From Exa’s public posts and my own search + napkin-math models to a concrete infra-cost picture.
🚀 I've been fortunate to work with top startups and research institutes from all over the world:
- Software Engineer, Core (Rust 🦀): Qdrant (Berlin)
- Software Engineer: SuperTokens (YC 20) (San Francisco)
- Software Engineer Intern: FamPay (YC 19) (Bangalore)
- Google Summer of Code: Software Heritage (Paris)
- ML + SDE Intern: Logy.AI (Hyderabad)
- ML Research Intern: IIT Mandi (Mandi)
Judging and Talks 🎤
I'm open to judging and speaking opportunities at reputed hackathons and conferences/meetups. Some of my past talks can be found at talks.kshivendu.dev.
For collaboration or a technical question about search and databases, email [email protected]. I'm also on Twitter and LinkedIn.