BUILDING Ideas into useful things

Building software
that makes a difference.

Hello, I’m Hiren Software Engineer

Paris, Ontario

Based in Ontario. Open to thoughtful collaboration across time zones.

Across time zones.

Hiren Thakore outdoors
ALWAYS CURIOUS · ALWAYS BUILDING

Software engineer.
Turning ideas into
useful things.

A little more beyond the code.

About me

THE TOOLS BEHIND THE IDEAS

Curiosity is part of my stack.

Cloud systems, AI agents, and the space in between.

Explore my toolkit
LET’S BUILD SOMETHING USEFUL

Your next idea.
Our next conversation.

MORE ABOUT ME

I’m Hiren, a
Software Engineer.

I’m a software engineer focused on backend systems, cloud infrastructure, and AI-powered products. I enjoy taking a complicated problem and turning it into reliable software that people can use.

My work spans Python, event-driven microservices, commerce integrations, and developer tooling. At Evertz, I build serverless systems and APIs; before that, I worked on cloud platforms and integrations at eHouse Studio and VL OMNI.

Outside that work, I’m exploring AI agents, model benchmarks, and tools that make engineering workflows easier to understand. Curiosity usually leads me to the next thing I build.

Skills & Tools.

A practical toolkit, shaped by the things I build.

Thoughtfully responsive.
Beautifully presented.

A closer look at ARA Calls, from desktop to pocket-sized screens.

ARA Calls homepage on a phone
Explore ARA Calls
CASE STUDIES

Featured Work.

All work
ARA Calls homepage on desktop
01AI RECEPTIONIST

ARA Calls

A callback-focused receptionist workspace for Canadian HVAC and plumbing businesses. Business facts and receptionist rules help collect caller details, while a shared workspace keeps callback requests ready for review.

  • Business facts and receptionist rules in one workspace.
  • Caller details and callback requests for human review.
  • Designed around Canadian HVAC and plumbing businesses.
ReactTypeScriptAI workflows
Visit the website
Conceptual project artwork: a precision optical timing instrument with three parallel signal paths
MCP & MODEL BENCHMARKINGLLM Speedtest
02MCP & MODEL BENCHMARKING

LLM Speedtest

An MCP server for comparing model inference speed from your own environment. Measure first-text latency, total response time, and estimated output throughput across explicitly selected models and configured providers.

  • Benchmark one model or an explicitly selected model set.
  • Compare first-text latency, response time, and throughput estimates.
  • Use your own provider accounts through an MCP client.
JavaScriptNode.jsMCP
Explore the code
agent-tallyOPEN SOURCE

~ pip install agent-tally

Know what your agents cost.

$ agent-tally --help

Track usage. Surface gaps.
Understand the bill.

03OPEN-SOURCE DEVELOPER TOOL

agent-tally

A local companion for understanding what AI coding agents cost. Wrap a CLI session, track detected usage and local cost estimates, and make missing usage or unknown model pricing visible rather than treating it as proof of a free run.

  • Wrap agent CLIs and track detected usage as they run.
  • Review local cost estimates and configure budget controls.
  • Surface missing usage and unpriced-model states.
PythonCLIAI agents
Explore the code
A LITTLE MORE TO EXPLORE

Follow the curiosity.
See what’s next.

The experiments, commits, and tools behind the work.

Open source

A good conversation
can start something.

Ideas, engineering questions, or a simple hello.

The tools and systems that shape how I build.

My toolkit