Senior Python Developer

Senior Python Developer New Street Technology builds AI-powered software platforms that transform how financial institutions operate. Established in 2017, we serve 170,000+ customers across lending, trade finance, and payments recognised by the Financial Times as the 7th fastest-growing fintech in Asia Pacific and a Deloitte Technology Fast 50 company. Department Engineering • Backend & AI/ML (MiFiX.ai team) Experience 47 years Reports to Engineering Manager / Head of Engineering Location Bengaluru Positions 1 Role code NST-SRD-001 About the Role You will own delivery across four areas — backend services (REST APIs and microservices for banking integrations and loan workflows), data and analytics (ETL pipelines, lending portfolio reporting), classical AI/ML (credit scoring, fraud detection, customer segmentation), and GenAI applications (LLM-powered features and agents built on top of the platform). The role spans design, build, ship, and mentor. This is a hands-on, senior IC role. You write the hard code, raise the bar on architecture and review, set the patterns juniors will copy, and engage with clients when needed. What You'll Do Design and build production services. FastAPI microservices for credit bureau, DEDUPE, Penny Drop, UPI, IMPS, loan disbursement and repayment workflows — including auth (OAuth, JWT), async tasks (Celery), and PostgreSQL schema design. Own the AI/ML integration layer. Take credit-risk scoring, fraud detection, and customersegmentation models from notebooks to production — Dockerised services, AWS SageMaker (or equivalent), monitoring, retraining triggers. Build GenAI applications. Ship LLM-powered features — RAG over policy or document corpora, structured extraction, agent flows, customer-support assistants — with proper prompt engineering, evaluation, and guardrails. You decide where an LLM earns its place and where deterministic code wins. Build data pipelines. ETL across our internal and partner sources; automated reports and dashboards for loan portfolio analysis and profitability metrics. Set the engineering bar. Pull-request reviews, refactors, testing standards, and the patterns the rest of the team copies.Mentor juniors. Pair with the two junior developers and grow them — without becoming a bottleneck. Engage with clients. Walk a bank engineer through an API contract or a debugging session; translate field issues into engineering fixes. Use AI assistants strategically. Cursor, Claude, Copilot — leverage them to compress build cycles, with the judgment to know what they will get wrong. What You Bring 4–7 years of backend engineering with Python in production. Something you built must be running live today — you can describe what it does, who depends on it, and how it is monitored. Strong FastAPI (or Flask/Django) with async/await fluency — you understand event loops, blocking I/O, and connection pooling. Solid PostgreSQL — schema design, indexes, query plans, transactions, and migrations under load. Comfort with SQLAlchemy or another ORM. Practical ML experience — at least one model you trained, evaluated, and shipped to production using scikit-learn, XGBoost, or equivalent. You can defend a model decision with numbers. GenAI application experience. You have built and shipped at least one LLM-powered feature or agent — RAG, structured extraction, or a tool-using agent. You know how to evaluate it, how to control its output, and how it fails. "I tried Claude in a side project" does not count; production exposure does. Data fluency — Pandas, NumPy, and ETL pipelines for real reporting use cases, not just exploratory notebooks. Comfort with Docker, Redis, Celery (or equivalent task queues), and Linux/Unix. Pragmatic system design — you can justify a service boundary, a queue, or a cache, not just sprinkle them. Clean reviewable code; type hints; tests that catch regressions, not just hit coverage targets. Bachelor's degree in CS, IT, Mathematics, or a related discipline. What Would Be Great to Have Fintech, banking, or lending domain experience — credit bureau integrations, UPI/IMPS, loan lifecycle. Deeper ML exposure — TensorFlow / PyTorch, deep-learning models, or NLP for document processing. AWS production experience — Lambda, S3, RDS, SageMaker — and exposure to message queues (RabbitMQ, Kafka).Hands-on with agent orchestration frameworks (LangChain, LangGraph, LlamaIndex, or custom orchestration). Multi-tenant SaaS patterns — tenant isolation, per-tenant data, control-plane / data-plane splits. Tech Stack Python: Python 3.11+, FastAPI / Flask / Django, Pydantic, SQLAlchemy Data & ML: Pandas, NumPy, scikit-learn, XGBoost, Matplotlib, Seaborn, Plotly, Jupyter GenAI: Claude / OpenAI / Mistral APIs, LangChain / LangGraph / LlamaIndex, RAG with pgvector / Qdrant, evaluation (RAGAS, DeepEval), prompt engineering Databases: PostgreSQL, Redis Infrastructure: Docker, Git, AWS (Lambda, S3, SageMaker), Celery, RabbitMQ Tools: VS Code / PyCharm, pytest, Postman