Oct 2026 – Ongoing
Intern
Onsite · Bengaluru
QA Intern·Evaratus
(formerly Scaler AI Labs)
- Foundry
- Loop
- GitHub
- Claude
- Docker
- Selenium
- Python
- Reinforcement learning
RL task validation for the MTS team
Validating reinforcement-learning tasks end to end on Foundry: checking task specs, expected outputs and grading logic so every task is solvable, unambiguous and scored fairly before it reaches training.
FoundryTask validationGradersEdge-case hunting in agent environments
Stress-testing RL environments with boundary and adversarial inputs to surface reward hacks, broken state transitions and failure modes, then flagging them with clear repro steps.
Edge casesReward designFailure modesAutomated browser checks with Selenium
Scripting Selenium runs to replay agent trajectories in web-based environments, turning manual click-through QA into repeatable checks.
SeleniumPythonTest automationReproducible environments & AI-assisted review
Running environments in Docker so bugs reproduce the same way every time, and using Claude to trace long agent runs and speed up review. Issues and fixes tracked through GitHub and Loop.
DockerClaudeGitHubLoop
