TypeScript Engineer, AI Coding Agent Evaluator - remote job

G2i is hiring a TypeScript Engineer, AI Coding Agent Evaluator. The role sits in IT and programming and is fully remote. The company places no restriction on where the candidate lives, so you can apply from anywhere.

The main tools named in the advert are: TypeScript, Python. Your CV should show concrete examples of exactly these skills.

The employer published the pay: $100 - $200 an hour. The advert sets no condition on working hours.

This vacancy passed an automated check: the list excludes any advert requiring a foreign work permit, visa sponsorship, a particular citizenship, or residence in a specific country.

Senior AI Interaction Evaluator (Codex / Claude Code)

Contract | $100-$200/hour | 10-20 hrs/week | Start ASAP (through early May)

Check out this Loom video for more details!

We want a deeply experienced software engineer (SR+) to judge the quality of interactions people have with current coding agents, including OpenAI Codex and Claude Code.

This is not a traditional engineering role. No production code gets written here. The harder question you answer: does the model reason the way a great engineer reasons?

What This Role Actually Is

You review how AI coding agents act in real situations, looking at:

* Whether the answer is sensible

* Whether the preamble and reasoning add value

* Whether the result shows sound engineering judgment

* Whether the exchange reads right to a seasoned developer

Engineering taste drives this work, not syntax checking.

What You'll Be Doing

* Review complete AI-generated coding interactions from start to finish

* Decide whether outputs are:

* Useful

* Correct (at a high level)

* In line with how a strong engineer would reason

* Rate explanations and reasoning, not code alone

* Separate quality tiers from one another (for instance, why one answer earns a 2 and another a 4)

* Give sharp, opinionated notes on:

* What worked

* What didn't

* What felt "off" or misleading

* Shape the definition of great when people work with tools like Cursor

What We Mean by "Taste"

The engineers we want can settle questions such as:

* Would a strong engineer really put it this way?

* Does the explanation actually help, or is it merely accurate?

* Is the model steering the user, or just dumping output?

* Does this exchange earn trust or lose it?

Subjective calls, made rigorously, should suit you.

Who You Are

* Staff / Principal-level engineer (or comparable experience)

* Solid grounding in one of these:

* TypeScript / JavaScript

* Python

* Practical use of:

* OpenAI Codex

* Claude Code

* Cursor

* Deep knowledge of current AI-assisted dev workflows

* Can judge code without running it or reading every line closely

* At ease delivering blunt, opinionated notes

* Holds a high standard for what counts as "good engineering"

Nice to Have

* Time spent in Cursor or other AI-first IDEs

* Earlier contact with prompt design or evaluation workflows

* Mentoring of senior engineers, or ownership of engineering standards

Engagement Details

* Rate: $100-$200/hour

* Hours: ~10-20 hours/week

* Duration: Through early May (extension possible)

* Start: ASAP

* Process:

* Take-home evaluation exercise

* One behavioral interview

Originally posted on Himalayas