Roger R. Thomas, B.Sc. (Hons), M.Sc.

• Location: Dublin, Ireland
• Website: rt.two4u.fun
• LinkedIn
• GitHub

Professional Summary

Software engineer with a background in astrophysics and scientific computing. Early work using C/C++ to develop scientific models led to more than a decade of building backend services, APIs, data platforms, and high-performance systems. My experience spans Python web frameworks including FastAPI, Django/DRF, and Flask; relational and NoSQL databases including MySQL, PostgreSQL, DynamoDB, and MongoDB; and the PyData ecosystem, including Pandas, Dask, NumPy, and Numba.

Currently a Document Intelligence Architect (Principal Engineer) at Nitro, building scalable document-processing systems using machine learning and generative AI, including document indexing, semantic retrieval, and retrieval-augmented generation (RAG).

Experience

Principal Engineer, Nitro, Inc. – Dublin, Ireland

May 2023 – present

• Design and build scalable document-intelligence services, combining hands-on backend engineering with system design across machine learning, generative AI, document processing, indexing, semantic retrieval, and retrieval-augmented generation.

• Technology

  High level:

  Python, FastAPI

  AWS:

  EKS, Lambda, DynamoDB, Textract, Step Functions, Bedrock

  Azure:

  OpenAI, AI Search

  Infrastructure and delivery:

  Terraform, GitHub Actions

Senior Lead Software Engineer, Oliver Wyman – Dublin, Ireland

Jan 2022 – May 2023

• Led platform architecture and engineering delivery, setting technical direction across API design, infrastructure modernisation, reliability, and feature development.

• Partnered with product owners, clients, and stakeholders to shape priorities, plan delivery, mentor engineers, and support the platform's long-term operation.

• Technology: Python, Django/DRF, Airflow, Argo Workflows, MongoDB, PostgreSQL, SQL Server, Docker, Kubernetes, and Azure.

Lead Software Engineer, Oliver Wyman – Dublin, Ireland

Jan 2019 – Jan 2022

• Led the technical design of the platform's analytics architecture, spanning Python, Airflow, Django/DRF, MongoDB, and the PyData ecosystem.

• Designed and delivered a general-purpose asynchronous job-execution system for distributed data processing using Python, Dask, and Azure.

Senior Software Developer, Oliver Wyman – Dublin, Ireland

June 2017 – Jan 2019

• Built and delivered core retail SaaS features across backend services, data processing, and computational workloads using Python, SQL Server, MongoDB, and the PyData ecosystem.

Technical Lead / Software Developer, Creme Global – Dublin, Ireland

Jan 2014 – June 2017

• Led and modernised the Python and Django backend for a computational risk-exposure platform, including REST APIs and automated testing.

Python Developer, AGOGO Amalgamated, Inc. – San Francisco, USA

Feb 2013 – Dec 2013

• Developed and maintained a large-scale Python web-scraping framework.

Mathematical Software Developer, Creme Global – Dublin, Ireland

Oct 2011 – Dec 2012

• Developed scientific models and distributed Python and AWS systems for consumer risk and chemical-migration analysis.

Selected Projects

TaglyDB

A simple REST database for tagged JSON and blob storage – no schema, just tag it and find it.

• Store JSON or blobs, tag them, and query by tag combinations – no tables to design.

• REST API with per-integration API keys and signed, shareable links.

• FE: Next.js(TS), Tailwind CSS, daisyUI.

• BE: Hono(TS), Cloudflare Workers, D1, Typebox.

jero

A high-performance, msgspec-first ASGI framework for modern Python.

• Class-based routing through Resource and Endpoint, with type hints as the API contract.

• Strict startup validation, typed request binding, typed responses, and explicit application wiring.

yeetr

A tiny, typed, signature-driven Python CLI runner.

• Runs ordinary synchronous or asynchronous Python functions as command-line tools.

• Uses function signatures for arguments and options, with rich help and support for structured types.

lothc

A typed HTTP client for Python, built on Rust-backed pyreqwest.

• Typed request and response decoding into pydantic models or msgspec structs, with sync and async clients sharing one API.

• Published benchmarks showing validated msgspec decoding outperforming unvalidated dict parsing.

two4u

A mobile-first sports prediction application built for a long-running match-day game with friends.

• FE: SvelteKit(TS), Tailwind CSS, Skeleton.

• BE: Python, FastAPI, DynamoDB.

• Infra: AWS (API Gateway, Lambda, S3), Cloudflare, Terraform.

Education

Trinity College Dublin, M.Sc. (Hons), High Performance Computing – Dublin, Ireland

2010 – 2011

University College Dublin, B.Sc. (Hons), Astrophysics – Dublin, Ireland

2006 – 2010

Technologies

Languages: Python (primary); TypeScript (working knowledge); C, C++, Go, Java (familiar)

Python: FastAPI, Django/DRF, Flask, ASGI, Pandas, NumPy, Numba, Dask

Cloud & Infrastructure: AWS, Azure, Kubernetes, Docker, Terraform, Argo Workflows, Airflow, GitHub Actions

Data: DynamoDB, MongoDB, PostgreSQL, SQL Server, MySQL, distributed data processing

Engineering: Backend engineering, API design, observability, CI/CD, technical leadership and mentoring