Mariusz – Node.js, React, Typescript

A prototype tool that automates title report generation for real estate by pulling public property and land records and turning them into structured, usable reports. It scrapes data from regional government sources and public APIs, then uses AI APIs to extract and summarize the relevant details into clean outputs like PDFs and structured JSON. I worked as the full-stack developer on the project, owning the backend end to end — building the scraping and data-access layer, integrating AI-based summarization, and generating the final report formats — while also contributing to overall project structure and a lightweight frontend where needed.

• Built a production AI title-analysis platform with LangChain, multi-provider LLMs (Google/OpenAI/Anthropic), and 30+ Inngest workflows, cutting large-order analysis from ~74 min to ~7.6 min for ~146 documents (~90% faster).
• Designed a multi-source AI document pipeline across 4 record types (title chain, PACER, name search, pre-1980), handling up to 350 docs/order with 50 parallel AI calls and a 90% success threshold, so high-volume orders complete without manual orchestration.
• Implemented LangGraph validation agents (20 nodes) with deterministic checks plus LLM judging (90% address / 88% city match, 70% pass threshold), reducing false rejections and routing ambiguous cases to structured manual review.
• Integrated AI outputs into product workflows via 11 checklist codes and findings, so failed/missing OCR docs are surfaced as retryable warnings instead of silently passing as 100% complete.
• Shipped an AI assistant with 12 tool-use functions over live title data, capped at 30 tool calls/turn and an 80K-char result budget, turning batch analysis into an interactive operator workflow.

Led discovery and implementation of bank details project that seamlessly captured financial information during client onboarding workflow, securely transmitted this data to our core accounting service via event-driven architecture, and established bidirectional synchronization with NetSuite ERP system that ensured real-time accuracy of payment information across all platforms while maintaining strict compliance with financial regulations

• Led the discovery and implementation of a bank details project to streamline financial operations and improve data accuracy;
• Assisted with integrating third-party ERP systems, including NetSuite and Microsoft Dynamics 365, to enable data synchronization across platforms;
• Participated in implementing serverless AWS solutions using Lambda and DynamoDB to support the team's cloud-native architecture goals.

This is an all-in-one platform that helps users create more sustainable packaging quickly and at scale. Users can design the packaging online in minutes or collaborate with experts to develop a bespoke solution.

• Designed and implemented a product configurator for e-commerce basket
• Architected and implemented an integrated stock-management system, unifying various spreadsheets and web apps into one comprehensive platform, driving 55% of company's revenue and optimizing operations;
• Spearheaded the creation of an innovative greenfield platform, designed and developed with micro-frontends to enable seamless team collaboration, which improved operational efficiency and client satisfaction, establishing the platform as the default offering for all client interactions.
• Implemented a cross-application passwordless and SSO solution, enhancing security and accessibility across all enterprise applications;
• Transformed the hiring process by refining interview practices and standardizing questions;
• Provided engineers with the support, guidance, and mentorship to advance their careers, resulting in the promotion of two engineers to senior positions.
• Established coding standards and best practices, conducting regular code reviews and knowledge-sharing sessions;
• Led several POCs on integrating LLMs to automate the extraction of data from supplier orders in PDF format;
• Led implementation of the supplier quotes uploader using Next.js and Postgres.

This is a platform for Real Estate Agents to gain unique access to leads based on signals coming from the market on who are willing to sell and establish custom outreach channels on LinkedIn on behalf of agents. It has involved custom API integrations with data sources (planning applications, land registry, company houses, and scraping property listings from public listing pages like rightmove and zoopla) and setting up MVP for agents to use and gain access to leads.

• Used LLM (OpenAI) for conversation classification and automated tagging;
• Designed and scaffolded MVP;
• Provided web scrapping solution for obtaining data from public sources on property listings;
• Integrated firebase auth with react SPA;
• Designed and scaffolded async flows in Inngest;
• Designed DB schema;
• Integrated application with different source providers (land registry, heyreach.io, overlead.io, LinkedIn, Slack, etc.);
• Configured authentication using Firebase
• implemented subscription payments with Stripe
• Set up CI/CD and deployed on Cloudrun on gcloud and Vercel;
• Added monitoring and observability (sentry.io).

Proximie is a technology platform that allows clinicians to virtually ‘scrub in’ to any operating room or cath lab from anywhere in the world. The target audience is surgeons, hospital coordinators, health professionals, and medical students. The app consists of 2 client-facing single-page apps: the dashboard app and the live app.

• Was responsible for leading a "dashboard" team of 6 software engineers to deliver the solution;
• Helped to grow the team members, organized show & tell sessions and OKRs reviews;
• Implemented an 0Auth invite-only flow integration and turned off the standard registration as only hospitals and unis were allowed to invite trusted users;
• Implemented 0Authauth management API integration to allow managing users and organizations;
• Wrote unit & integration tests;
• Implemented cypress e2e tests, including 0Auth login flow + user journeys;
• Improved Devex, including local development: DB seeds, CLI commands, and debugging solutions.

Main achievement:

• As this industry was highly regulated by HIPAA compliant, data had to be stored in the location where the existing session/organization, so Mariusz was involved in designing multi-region data center architecture for providing a de-identified data storage solution - imagine a central server hosting non-PII in central location and store reference to data in data stores in satellite locations, the challenge was to marry up those data in the dashboard and get accurate PII in the runtime performance, and fallbacks.

Nationwide Building Society is a building society owned by its members. Anyone who banks, saves, or has a mortgage with them. They run for members' benefit and help the communities around them. The app Mariusz worked under was for mortgage brokers to make the process of streamlining mortgages easier, including all its stages and integration with banking services using event-driven architecture.

Main responsibilities:

• Developed several features of mortgage application single page application, using React.js, React-Router v4, Context API, SignalR, Axios, Formik, Semantic-UI, NBS-UI Library, Jest;
• Worked effectively with UI/UX designers producing high-quality prototypes using storybook and invasion;
• Created and maintained shared components and themes of internal UI library (NBS-UI), using React.js, Semantic-UI, Storybook, Jest
• Helped develop node.js CLI tool to speed up BDD testing;
• Assisted in Jenkins pipelines setup, maintained and monitored Kubernetes microservices using EFK stack;
• Worked in cooperation with .NET engineers pioneering implementation of Event Driven Architecture at Nationwide Building Society;
• Mentored Technology Development Programme graduates, enabling them to gain real-world experience.
• Worked on C#/.net microservices (affordability-checker, eligibility-checker, message-service, api gateway)

Main accomplishment:

• Worked on first market mortgage product for over 55s from service design to live in 10 months;

HT2Labs help learning and development get brighter through data. Mariusz developed and maintained the GDPR application, which allowed clients to stay compliant with EU law regulations.

• Created front-end and backend architecture;
• Dockerized solution;
• Implemented CI/CD pipelines;
• Set up infrastructure, including MYSQL DB;
• Wrote unit, integration, and e2e tests.

Mariusz has created a multi-threaded CLI tool in Java to import 16GB CSV files (Experian data) into a local database used for marketing purposes.

• Implemented CLI app using java and 4 threads concurrency;
• Designed and implemented DB schema (migrations)
• Maintained the app over time according to the new requirements