Senior Software Engineer, ML Platform

Department: Engineering

Location: San Francisco, CA

Employment Type: FullTime

About Us:

At Parafin, we’re on a mission to grow small businesses.

Small businesses are the backbone of our economy, but traditional banks often don’t have their backs. We build tech that makes it simple for small businesses to access the financial tools they need through the platforms they already sell on.

We partner with companies like DoorDash, Amazon, Worldpay, and Mindbody to offer fast and flexible funding, spend management, and savings tools to their small business users via a simple integration. Parafin takes on all the complexity of capital markets, underwriting, servicing, compliance, and customer service for our partners.

We’re a tight-knit team of innovators hailing from Stripe, Square, Plaid, Coinbase, Robinhood, CERN, and more — all united by a passion for building tools that help small businesses succeed. Parafin is backed by prominent venture capitalists including GIC, Notable Capital, Redpoint Ventures, Ribbit Capital, and Thrive Capital. Parafin is a Series C company, and we have raised more than $194M in equity and $340M in debt facilities.

Join us in creating a future where every small business has the financial tools they need.

About The Position

We’re looking for a software engineer to join Parafin’s Infrastructure team and lead the evolution of our ML Platform. This role is critical to building reliable, scalable, and developer-friendly systems for model experimentation, training, evaluation, inference, and retraining that power underwriting and other ML-driven products for small businesses.

As a Software Engineer, you’ll design, build, and maintain the core abstractions and platforms that let data scientists ship high-quality models to production—safely and quickly. You’ll partner closely with Data Science and Platform Engineering, own the ML platform end-to-end, and develop batch and real-time underwriting infrastructure.

What You'll Do

• Turn notebooks into software. Decompose data scientist training/inference notebooks into reusable, tested components (libraries, pipelines, templates) with clear interfaces and documentation.

• Create developer-friendly ML abstractions. Build SDKs, CLIs, and templates that make it simple to define features, train/evaluate models, and deploy to batch or real-time targets with minimal boilerplate.

• Build our real-time ML inference platform. Stand up and scale low-latency model serving.

• Expand batch ML inference. Improve scheduling, parallelism, cost controls, observability, and failure/rollback for large-scale batch scoring and post-processing.

• Own and expand the feature store. Design offline/online feature definitions, high read/write throughput, and consistent offline/online semantics.

• Platform reliability and observability. Instrument training/inference for latency, throughput, accuracy, drift, data quality, and cost; build alerting and dashboards; drive incident response and postmortems.

• Underwriting infrastructure partnership. Support production batch and real-time underwriting systems in collaboration with Data Science; collaborate on model interfaces, SLAs, safety checks, and product integrations.

What We Are Looking For

• 5+ years of software engineering experience, including experience on ML platform/MLOps systems (training, deployment, and/or feature pipelines).

• Strong Python; solid software design and testing fundamentals. Proficiency with SQL; hands-on Spark/PySpark experience.

• Knowledge of ML fundamentals—probability & statistics, supervised vs. unsupervised learning, bias/variance & regularization, feature engineering, model evaluation metrics, validation strategies, and production concerns like drift, stability, and monitoring.

• Expertise with modern data/ML stacks—AWS, Databricks (workflows, lakehouse, MLflow/registry, Model Serving), and Airflow (or equivalent orchestration).

• Experience building real-time systems (service design, caching, rate limiting, backpressure) and batch pipelines at scale.

• Practical knowledge of feature-store concepts (offline/online stores, backfills, point-in-time correctness), model registries, experiment tracking, and evaluation frameworks.

• Strong problem-solving skills and a proactive attitude toward ownership and platform health.

• Excellent communication and collaboration skills, especially in cross-functional settings.

Bonus Points

• Databricks experience (MLflow, Model Serving).

• Experience with feature stores (e.g., Tecton, Feast) and streaming (Kafka/Kinesis).

• Experience with fintech, risk, or underwriting systems; familiarity with model safety checks, rejection/override flows, and auditability.

• Background with A/B testing platforms, shadow/canary deployments, and automated rollback.

• Experience with low-latency inference systems.

What We Offer

• Salary Range: $230k - $265k

• Equity grant

• Medical, dental & vision insurance

• Work from home flexibility

• Unlimited PTO

• Commuter benefits

• Free lunches

• Paid parental leave

• 401(k)

• Employee assistance program

If you require reasonable accommodation in completing this application, interviewing, completing any pre-employment testing, or otherwise participating in the employee selection process, please contact us.