Hiren Thakore

Software engineer building AI agents, developer tools, and automation-heavy products.

I like systems that do real work: agents with memory, tools that remove busywork, workflows that ship without drama, and small products that prove an idea fast.

Portfolio · LinkedIn · X · Email

---

What I care about

• Software engineering: clean systems, useful abstractions, good DX, boring reliability.
• AI agents: local-first assistants, tool use, memory, scheduled work, subagents, agent security.
• AI workflows: turning repeatable work into automation, from research to code review to shipping.
• Developer products: small tools that help builders move faster without adding process.
• Open source: learning in public, contributing where I can, and studying how serious software gets made.

Current direction

focus:
  - AI agents that can operate across files, browsers, terminals, and chat
  - developer tools for API work, docs, evals, prompts, and launch workflows
  - micro-SaaS experiments built fast and tested in the open
  - automation pipelines that turn ideas into shipped artifacts
  - practical AI engineering: cost, evals, security, reliability, UX

Projects I am building around

• llm-speedtest-mcp
  Benchmark AI model inference speed, like speedtest.net for LLMs. MCP server, zero telemetry.

• hermes-exploration-plugin
  Exploration layer for AI agents to discover tools, APIs, models, and workflows.

• agent-tally
  Cost tracking across AI coding agent CLIs.

• apicaller
  Paste an API URL and get the curl commands you need.

• vibecheck
  Audit AI-generated code for security issues, weak patterns, and maintainability problems.

• realestate-cmo
  Vertical AI CMO experiment for real estate operators.

• promptforge · readmeforge · ai-citation-monitor
  Small builder tools for prompt work, documentation, and AI search visibility.

Open source rabbit holes

I spend a lot of time studying and forking projects in:

• agent frameworks and personal assistants
• MCP servers and tool registries
• LLM gateways, evals, and observability
• ML infrastructure and workflow orchestration
• Python developer tooling

A few areas I keep coming back to: agents, MCP, LLM APIs, evals, automation, workflow engines, developer experience, AI security.

Stack I reach for

Languages: Python, TypeScript, JavaScript, Go, Solidity
Frontend: React, Next.js, Tailwind, Astro
Backend: FastAPI, Node.js, Django, REST, GraphQL
Data/infra: PostgreSQL, Redis, MongoDB, Docker, GitHub Actions, AWS
AI/dev workflow: MCP, agent tool use, browser automation, evals, LLM APIs, prompt systems

GitHub stats

How I think about engineering

I like tools that are small enough to understand and useful enough to keep using.

The best AI workflow is not a flashy demo. It is the boring loop that runs every day, catches errors, saves time, and leaves a trail you can debug later.

That is the kind of software I want to build more of.

---

Building in public. Mostly software, agents, automation, and experiments that ship.