Narasimha Badrinath

I build AI agents that survive the real world.

I'm a Staff Developer Advocate at DigitalOcean. I build AI apps and agents, contribute to the open-source tools developers use to ship them, and write about what I learn.

Selected work

Voice assistant for macOS

Nix

Nix lives in my Mac's menu bar. I say "Hey Nix", ask something, and it answers out loud. The thinking is done by whichever coding agent I'm signed in to (Claude Code, Codex or Antigravity), so it can read files, run commands and use my tools, within permissions I set. Speech recognition runs on the Mac with Whisper, five times faster than a hosted service (265ms against 1.3s), and the audio never leaves the machine.

500+ commits · Swift app with a local Node service

Official integration

langchain-gradient

I wrote DigitalOcean's official LangChain integration from the first commit, and I maintain it. LangChain apps use it to call models on Gradient serverless inference, and it streams responses and reports token usage. I added LangChain 1.x support, reviewed and released the 1.0 move to the new pydo SDK before the old SDK retired, and I review and merge community pull requests, such as the one that added tool calling.

18,000+ downloads · Python, on PyPI

$ pip install langchain-gradient

from langchain_gradient import ChatGradient

llm = ChatGradient(model="llama3.3-70b-instruct")

for chunk in llm.stream("What is RAG?"):
    print(chunk.content, end="")

Benchmark

Agent Memory Showdown

I benchmarked six agent memory backends over 30-turn conversations, scored by a judge from a different model family. Buffer, summary and Mem0 recalled everything. Qdrant, pgvector and the LangGraph store kept 87 to 95%, at about half the cost of keeping the full history, and every one of their misses was a fact that had been corrected. The rolling summary cost 4.8 times the full history.

3 runs per backend · Claude Sonnet 5.5, judged by GPT-6 Sol, MiniLM embeddings

Recall over 3 runs and cost per 1,000 conversations. The recall intervals overlap, so this run ranks cost, not recall. Highlighted rows cost least.
BackendRecallCost
Buffer100%$391
Mem0100%$963
pgvector (lowest cost)95%$193
Qdrant (lowest cost)89%$212
LangGraph store (lowest cost)87%$196
Summary100%$1,862

Architecture gallery

AI systems I've built

Five AI systems shown in depth: the design decisions, the trade-offs and the measured results. Eight supporting builds follow. Each one is drawn as its architecture. Open one to see the full diagram, the design notes and, where it is public, the code.

Open source I maintain and contribute to

  • llama-index-retrievers-digitalocean-gradientai The matching LlamaIndex retriever integration for Gradient, for RAG on serverless inference.
  • ai-ml-bootstrapper Hands-on demos and best practices for running AI/ML workloads on DigitalOcean. Top contributor; 26 stars.
  • GuardianAI My MCP gateway that applies a YAML policy to agent tool calls. It can deny a call, hold it for approval, or redact the result. Has Python and TypeScript SDKs and a live demo.

Latest writing

Live apps and side projects

AI projects

  • ArchitectAI LiveDescribe an app and watch agents plan, generate and check a cloud architecture for it.
  • Dead Startups Roast LivePick a failed startup, get it roasted by AI, then see how you'd rebuild it today.
  • Voice Course StudioAgents turn a raw recording into a finished explainer video, stopping three times for my review.

iOS apps

  • DailyBrickOffline habit tracker with streaks and widgets. No accounts, no subscription.
  • Silly DebatesPeople argue absurd topics for 24 hours, then AI reads every opinion and picks a winner.
  • FlipItCard-matching game with five themed decks and global leaderboards.

Speaking

Narasimha Badrinath on stage, presenting on running AI/ML workloads on DigitalOcean
Running AI/ML workloads on DigitalOcean

I give talks and run hands-on workshops on AI agents and taking LLM apps to production. I also review agent architectures for a few teams.

I'm always up for a meetup, conference or workshop. The easiest way to reach out is below.

Let's connect.

I'm always happy to talk about AI agents, developer tools, talks and workshops. Book a time, or pick whichever channel suits you.