We are an actively developing FinTech startup creating a consumer product for the international market. Our product analyzes user banking transactions, identifies regular payments and subscriptions, and helps find opportunities for expense optimization.

Currently, part of the analysis is performed manually. Our next task is to understand and build the correct automation architecture: where to use rules, where to use classic ML, where to use LLM, and how to combine them into a reliable production system.

We are looking for an Applied ML / LLM Engineer who can not just work with APIs of ready-made models, but solve applied problems with real data: design architecture, choose approaches, build ML/LLM pipelines, and measure the quality of the result.

What needs to be done: — Design a banking transaction analysis system: from raw data to structured results and user recommendations — Solve problems of transaction classification, merchant normalization, recurring payments detection, and anomaly detection — Determine which parts of the system are more effectively solved through deterministic rules, classic ML, embeddings, or LLM — Identify subscriptions and regular payments, billing frequency, and cost changes — Normalize merchants: understand that different transaction names refer to the same company/service — Separate real subscriptions from rent, transfers, returns, duplicates, and other regular transactions — Work with confidence scores and false positives/false negatives, define thresholds and logic for handling edge cases — Design the recommendation layer: how to find real savings opportunities for the user based on transaction history — Design the path from prototype to production: data, pipelines, quality monitoring, error handling, cost, and scaling

Requirements: — Practical experience with classification, recommendation systems, anomaly detection, entity/merchant normalization, or similar tasks — Ability to independently choose an architecture for a task and justify the choice of model/approach — Understanding of how production ML / LLM pipelines are built, not just prototypes and API integrations — Ability to define quality metrics, work with precision/recall, false positives, and confidence scores — Understanding of data requirements: what data is needed for training and validation, how to form datasets and ground truth — Ability to estimate timelines, resources, and major technical risks of moving a solution to production — Confident work with Python and the modern ML/LLM stack

Will be a strong plus: experience with FinTech / banking data / Open Banking / transaction categorization / recurring payments / recommendation systems, as well as real experience launching ML systems in production.

Conditions: — Work in a progressive fintech startup without bureaucracy, with a high speed of hypothesis testing and freedom of speech — Salary from $2500, base salary and KPI/OKR system — Work schedule: 5/2 or 6/1 by agreement — Fully remote employment