Atlas AI
Website:
helloatlas.in
Job details:
Atlas AI is an agentic AI platform built for Indian banks and NBFCs. We automate the manual, exception-heavy operations inside banking teams — onboarding, credit, and compliance — and we're live in production with some of India's top financial institutions.
What We're Looking For
- Final year student or recent graduate in CS, Statistics, or a related field.
- Strong proficiency in Python — not familiar with, proficient.
- Some exposure to LLMs — called an API, experimented with prompts, read the papers. Curiosity matters more than depth here.
- Solid understanding of ML evaluation — precision, recall, confidence calibration — in a practical, not just academic, context.
- Comfortable working with messy, real-world data and ambiguous problem statements.
- Familiarity with eval frameworks (Promptfoo, Arize, or custom Python pipelines) is a plus.
- Bonus: Exposure to OCR, document AI, or multimodal models.
What You'll Do
- Build and maintain eval pipelines for our document intelligence and agentic workflows.
- Construct labelled ground truth datasets across document types, regional languages, and edge cases.
- Run structured experiments on prompt strategies, extraction accuracy, and model confidence.
- Analyze model outputs at scale — identify failure patterns and work with the AI team to resolve them.
- Contribute to building our agentic onboarding product with real production data.
Perks
- Work directly with our Chief of AI on a live, production-scale BFSI system.
- Pre-PPO track — strong performers convert to full-time with equity.
- Stipend commensurate with a high-growth preseed startup.
Click on Apply to know more.