Regulated data, delivered in production.

AI tooling under strict data-protection rules

The challenge

Bring AI into development and operations at a bank, while rolling out Microsoft 365 Copilot and Claude under strict rules for protecting data.

What we did

Designed and shipped LLM integrations with OpenAI and LangChain, local-model inference with Ollama, and MCP-based tool access with FastMCP. Introduced the tools in stages, starting with AI-assisted code generation and review, and coached the team on where AI helps and where it only adds overhead.

Stack: Python, FastAPI, OpenAI, LangChain, Ollama, FastMCP

A data platform and API gateway for core banking

The challenge

Mission-critical banking applications needed a dependable data platform, and the CRM needed to connect to the core banking system without a tangle of point-to-point integrations.

What we did

Led architecture for the enterprise data platform (99.9% uptime) and its ETL integrations. Led the technical build of an API gateway on .NET 9 with clean architecture and domain-driven design, exposing five business domains through a single GraphQL interface, with OpenTelemetry tracing and structured logging. Rebuilt CI/CD and infrastructure as code.

Stack: Python, FastAPI, C# / .NET 9, GraphQL, PostgreSQL, Docker, Jenkins, OpenTelemetry

A data-exchange platform for public-health reporting

The challenge

Public-health data arrives in many formats and has to be ingested, validated and routed reliably across agencies and systems.

What we did

Built a high-throughput file upload service in Go (Tus protocol) and an Azure Functions event pipeline ingesting HL7, FHIR, CDA, XML and CSV. Designed a versioned JSON Schema validation system used across multiple CDC systems, and added observability and structured error handling. Migrated a core service from C# to Go.

Stack: Go, C#, Azure Functions, Event Hub, Terraform, GitHub Actions

Modernizing a national compliance system

The challenge

A long-running legacy compliance system for the Mammography Quality Standards Act (MQSA) program needed to become a secure, cloud-native platform for facilities nationwide.

What we did

Led a cross-functional team of 8 to 12 engineers as tech lead. Built the Angular and TypeScript front end, a RESTful Python API platform with Okta JWT authentication for a multi-tenant government application, a containerized delivery pipeline on AWS, and Lambda-based ETL that migrated data out of Lotus Notes, Access and Excel.

Stack: TypeScript, Angular, Python, SQLAlchemy, AWS (Lambda, S3, ECS/Fargate, RDS), Docker, Okta

A HIPAA-ready data foundation for a mobile platform

The challenge

An early-stage sports and health technology company needed a v1.0 mobile platform and a data model that could later support sensitive health data and AI.

What we did

Working directly with the founder, architected the platform end to end: a React Native app, a FastAPI backend, Redis and PostgreSQL. Implemented Auth0 authentication with PKCE and refresh-token rotation, an offline-first voice dictation pipeline, and the design of a HIPAA-ready health-data subsystem. The AI and agent layer is designed and planned; it is not yet in production.

Stack: TypeScript, React Native, Python, FastAPI, PostgreSQL, Redis, Auth0

Modern systems on small budgets

The challenge

Nonprofit and education clients needed dependable systems without enterprise budgets.

What we did

Rebuilt a nonprofit's website and content platform as a custom React, Express and PostgreSQL application with automated deployment, and moved hosting from WordPress to a self-managed server. For an early-childhood education organization, designed a serverless AWS backend, analytics tooling and infrastructure as code.

Stack: React, Node.js, PostgreSQL, Nginx, AWS Lambda, Terraform, CloudFormation