AI/ML Engineer Intern (Research & Development)

About the Role

Polarisk builds AI for financial crime compliance at banks. We are an early-stage team.

Key Responsibilities

• Build and improve the agent harness for compliance investigation
• Develop and evaluate LLM-based extraction, classification, and reasoning components
• Build synthetic data generation for agent and control testing
• Write evaluation harnesses that measure accuracy and reasoning quality
• Experiment with model training and fine-tuning where the evaluation data supports it
• Track agentic AI research (context management, long-term memory, agent learning) and bring what works into the product
• Document agent behaviour, evaluation results, and design decisions

Requirements & Tech Stack

• Strong Python and comfort with ML libraries
• Hands-on exposure to, or serious curiosity about, LLMs and agentic frameworks
• Ability to own a problem end to end, not only a sub-task
• Clear written communication; you will document and present your work
• Coursework or projects in ML, NLP, or distributed systems is a plus
• No fintech or compliance background needed.

Stack you will work with:

• LangGraph
• Claude
• GPT
• AWS Bedrock
• FastAPI
• PostgreSQL with pgvector
• MLflow

Qualifications

• Final-year students, recent graduates, and early-career engineers (including those with 2–3 years of experience) who are happy to join as an intern
• Able to work on-site, full-time, for the full 6 months

Internship Details

• Duration: 6 months
• Location: On-site, Whitefield, Bengaluru
• Paid internship from day one
• Start: Immediate joiners preferred; up to 1 month notice is fine
• Openings: 2