About PayPay and Pay 2 Development Center

PayPay, a fintech company providing a service enjoyed by over 60 million users merely 5 years since its launch in 2018 in Japan. The company is now home to a very diverse team of members from more than 50 countries. We grew to a team of several thousand employees in Japan but are far from over. We are still in the Day 1. Every day, new members join us from all over the world to create new value and deliver it to society.

Why India ?

To build our Payment services, we allied with Paytm, the biggest payment service company in India. Based on their customer-first technologies, we created and expanded the smartphone payment service in Japan. Therefore, we have decided to establish a development base in India, as we have been developing our services under the influence of India, a technological powerhouse from the beginning. We create user-first services with development members in Japan and Canada while pursuing the latest technology. That is what we will do here.

Our biggest competitor is "cash". We are seeking for people who can accept this challenge positively, brush up the product at a tremendous speed that other companies could never achieve, and who are passionate about promoting and spreading such a financial life platform in a short time along with professionalism.

Job Description

We are seeking an experienced Machine Learning Engineer with a strong background in Natural Language Processing (NLP) and building end-to-end ML systems. You should have deep expertise in information retrieval/search, vector databases and LLMOps.

In the near term, you will lead production-grade RAG (Retrieval-Augmented Generation) services that power customer support at scale - driving accuracy, latency, and coverage across the entire pipeline, from ingestion and indexing through retrieval, reranking, and LLM orchestration.

This role will expand to core AI initiatives for Financial Solutions to build ML/LLM Systems in alignment with our future roadmap.

Responsibilities

Own the RAG pipeline end to end for our customer support initiatives.
Systematically improve retrieval accuracy by productionizing techniques like:

Advanced chunking strategies for support documents and knowledge bases.
Embedding model fine-tuning on our domain-specific data.
Hybrid search (semantic + keyword) and implementing reranker models.

Build and maintain robust evaluation frameworks to measure system performance and guide improvements.
Build guardrails with SLMs to control faithfulness, answerability and reduce hallucination rate.
Ensure Safety and Governance for LLMs used through PII redaction, policy/guardrails integration, grounded citations and incident playbooks.
Collaborate with backend and data engineers to integrate ML models and search systems into our production environment.
In the future, design and implement other ML models for financial solutions, collaborating with data scientists working on different products.

Qualifications

6+ years of professional experience in machine learning engineering with a proven track record of building, deploying, and optimizing end-to-end AI/NLP systems in production
Educational background in Computer Science, Engineering, Mathematics, or a related field.
Familiar with LLM orchestration (LangChain/LlamaIndex or in-house), prompt/version management, tool use, and response streaming.
Core NLP/RAG Skills:

Deep understanding of Natural Language Processing (NLP), including Transformer architectures (e.g., BERT, Sentence-Transformers) and embedding techniques.
Hands-on experience with Vector Databases and/or semantic search technologies.
Expertise in ML frameworks like Hugging Face Transformers, PyTorch, or TensorFlow.
Experience with RAG frameworks like LangChain or LlamaIndex is strongly preferred.
Practical experience with evaluation - Golden Sets, A/B testing and RAG specific metrics

General & Financial ML Skills:

Good understanding of supervised and unsupervised learning and ensemble methods.
Proficiency with libraries like scikit-learn and XGBoost.
Experience in the FinTech industry or with financial/credit data is a significant plus.

Core Tech Stack:

Proficiency in Python. (Java/Scala is a plus).
Strong knowledge of database management systems (e.g., MySQL, PostgreSQL), ETL processes, and SQL.
Experience with MLOps practices and tools like Docker, Kubernetes, and cloud platforms (AWS preferred).
Familiarity with tools like Apache Spark.

Nice to Have

Multilingual Information Retrieval (especially JP/EN), domain ontology/KB design, schema/metadata strategies.
Safety/guardrails experience, privacy-by-design practices.
Experience with tooling like FastAPI/gRPC, TorchServe and monitoring with OpenTelemetry/Victoria Metrics.
Experience with Eval frameworks like Ragas, TruLens, DeepEval, LangSmith.

Expected Personality

Strong problem-solving and analytical skills.
Excellent communication and collaboration abilities.
Adaptability and a willingness to learn new technologies and techniques.
Proactive mindset with the ability to think strategically about system improvements.
Ability to make suggestions and improvements independently.
Logical communicator with the ability to coordinate smoothly with stakeholders.

Please refer PayPay 5 senses to learn what we value at work.