Kumar Shivendu

Software Engineer
DevTools & Information Retrieval

Hey there. I am Kumar Shivendu. 👨‍💻 Here's a PDF version of my Resume

I build auth devtools at SuperTokens (YC 20) and have previously worked at FamPay (YC 19). I'm proficient in Python and JS and have dabbled in production grade Java and Go as well. I'm passionate about devtools, information retrieval (search engines, query languages, LLMs, and databases), and open source.

These days I'm exploring the Generative AI space (especially semantic search and LLMs) by building projects like these on the weekends:

  • Benchmarking popular vector DBs - Qdrant vs Pgvector. I also published a 1M OpenAI embeddings dataset in the process. My work went viral on Twitter and was noticed by creators of Github Copilot, Yahoo search, and Spotify recommendation engine!
  • Suggest-AI 2k+ people used my side project to search for 10k+ different things. Eventually took it down because it didn't have a monetization strategy.
  • Cloning voices to make popular singers sing different songs, etc.

🚀 Experience:

🦄 I'm open to Software Engineering and DevRel roles at high growth startups in the devtools and the information retrieval space :)

Some of the cool things I shipped recently:

  • Monitoring my life: I'm collecting my health, time, and productivity data from all of my devices and putting it all together in Grafana dashboards. See /blog/quantified-self
  • Advanced search (Software Heritage): I designed advanced search features with a custom query language that operates on a corpus of over 250 million repositories containing more than 15 billion files. This archive is the largest software archive globally, surpassing even Github.
  • Multi-Factor Authentication (SuperTokens): I architected our APIs and SDK features that enable the addition of multiple login factors to any application in minutes. This was one of our most requested enterprise features and there's no other provider at the moment that offers the same level of flexibility.
  • Search Copilot (Personal project): I developed a browser extension that enhances the existing search capabilities of any website. It allows users to go beyond traditional filters and sorting methods, enabling them to ask questions in natural language. The extension utilizes intelligent heuristics for data scraping on the fly and employs vector embeddings for finding relevant items.

Reach out to me on Twitter, LinkedIn, or mail hello@kshivendu.dev if you need my help