Description

Job Title: Software Engineer – Full Stack
Location: Remote

Duration: 1 months contract - Extendable

Project Overview

We're building high-quality evaluation and training datasets to improve how Large Language Models (LLMs) interact with realistic software engineering tasks. You will have the opportunity to work on a diverse range of projects from helping models traverse complex code bases to building agents that improve model performance.

Role Overview — What Does a Typical Day Look Like?

Work across multiple different projects to improve LLM performance on code: sample projects

  • Leading and delivering end-to-end agent use cases such as home automation agents, coding copilots, or creative design assistants. 
  • Collaborate with the team to identify  edge cases and ambiguities in model behavior.
  • Review and compare 3–4 model-generated code responses per task using a structured ranking system.
  • Evaluate code diffs for correctness, code quality, style, and efficiency. Provide clear, detailed rationales explaining the reasoning behind each ranking decision.

 

Required Skills & Experience

  • Several years of software engineering experience, including 2+ continuous years at a top-tier product company (e.g., Google, Stripe, Amazon, Apple, Meta, Netflix, Microsoft, Datadog, Dropbox, Shopify, PayPal, IBM Research).
  • Strong expertise in building full-stack applications and deploying scalable, production-grade software using modern languages and tools.
  • Deep understanding of software architecture, design, development, debugging, and code quality/review assessment.
  • Proven ability to review code diffs and evaluate correctness, maintainability, and efficiency.
  • Excellent oral and written communication skills for clear, structured evaluation rationales.

Engagement Details

  • Commitment: flexible engagement, minimum 10 hrs/week, up to 40 hrs/week (partial PST overlap required).
  • Type: Contractor (no medical/paid leave).
  • Duration: 1 month potential extensions based on performance and fit.