Software Engineer - AI Evals and Benchmarking at Harmoney (W22) ₹2M - ₹3M INR • 0.30% Agentic AI solution for fixed income, credit and portfolio analysis Mumbai, MH, IN / Mumbai, Maharashtra, IN Full-time US citizenship/visa not required 1+ years About Harmoney We are building agentic AI for fixed income research, portfolio optimization, and credit analysis. We not only make financial analysis faster and cost-effective but also unlock things that could not be done manually before. Our 30-person team is mainly comprised of engineers and sales, and we work together to figure out user problems in the financial analysis space that can be solved using AI. About the role Skills: Python About us Our mission is to empower researchers, traders, asset managers and risk managers to make better decisions in financial markets Harmoney has built an agentic solution to deliver data and analytics to the users. This helps them generate insights in seconds vs hours with traditional means. Harmoney is backed by Y Combinator and has raised $5 mn in funding. Solid team - worked at Goldman Sachs, Amazon, Tata Capital, Paysense, and leading schools such as IITs and Purdue We are a product company, and our engineering team is at its heart. Our engineers thrive on innovation and the freedom to try out new things. Responsibilities We’re on a mission to make financial data more actionable. Looking for a Software Engineer who’s excited to build an evaluation framework. The software engineer will own the evaluation framework and will be responsible for ensuring quality control as we scale What will you do? Design and maintain the evaluation framework for agentic workflows Create and own datasets/ground truth (programmatic + human-in-the-loop), and manage drift detection Work closely with business teams to understand real-world cases to build a relevant evaluation framework What will you bring? Strong Python + SQL skills Curiousity to explore the AI tools early and iterate fast Why join Harmoney? Small, fast-moving team → lots of ownership Real impact at the intersection of finance + tech Space to experiment and grow with us Technology Google ADK for our agentic AI with continuous changes/improvements/additions Airflow and Kafka for datafeeds (we can sometimes add a new datasource in a day!) Evals using Langfuse or Arize Phoenix Everyone in the team uses Claude Code, Codex, or other tools. The team helps each other extract the maximum from AI - we love talking about work that is 90% AI developed and 100% human reviewed. Interview Process The interview process is designed to evaluate your ability and cultural fit at Harmoney. We prefer hiring engineers who can execute their work independently and have strong problem-solving skills. We have one phone screener followed by 2-3 technical rounds of interviews. The phone screener typically lasts 15 minutes and will give you an overview of Harmoney and ensure that your career goals are aligned with what we are looking for.