Mateusz Majchrzak
Senior Software Engineer
Experience
Senior Software Engineer
Ramp Network
- Led the architecture and production deployment of deep learning NLP models, applying transformer architectures for voice-based assistants in a fintech environment.
- Built semantic search pipelines using vector embeddings and retrieval-augmented generation (RAG) to improve support automation across user-facing platforms.
- Designed and monitored evaluation systems for bot behavior, integrating custom interpretability modules and real-time dashboards for KPI tracking.
- Applied MLOps best practices, integrating with GitHub Actions, Docker, and Terraform for CI/CD deployment of ML microservices.
- Implemented monitoring frameworks using Prometheus and Grafana to ensure model stability and detect drift.
- Translated research insights into performant, reproducible codebases using Jupyter Notebooks, Pytest, and LangChain.
- Fine-tuned open-source LLMs via Hugging Face Transformers to handle ambiguous queries with context-aware fallback logic.
- Championed model reproducibility, containerizing pipelines for retraining and rollback procedures across cloud stages.
- Integrated multilingual support through dynamic prompt injection and real-time language detection pipelines.
- Mentored junior developers on prompt engineering, evaluation strategies, and reliability-first ML system design.
Senior Software Engineer
C&F
- Spearheaded the development of a chat-based assistant for an NFT trading platform, optimizing NLP inference latency and accuracy.
- Deployed and maintained Python-based ML services built with FastAPI and React, orchestrated via Docker Compose and AWS Lambda.
- Developed a real-time model feedback loop, collecting user interactions to evaluate precision and recall metrics.
- Leveraged Watson Assistant and Google NLP APIs to automate support workflows, reducing human intervention by 25%.
- Built internal model comparison dashboards for evaluation against baseline QA systems using React and TailwindCSS.
- Implemented automated testing suites with Cypress, Pytest, and ESLint to ensure system integrity pre-deployment.
- Established cloud-first CI/CD ML workflows with Terraform and AWS ECS, reducing model rollout friction by 40%.
- Introduced embedding similarity matching for real-time query ranking, reducing irrelevant results by 60%.
- Developed and documented agent-based tools using OpenAI APIs, enabling multi-turn dialog strategies.
- Applied FAISS and Pinecone for rapid vector search across knowledge bases, improving retrieval time for support agents.
Software Engineer
IMAGINATION
- Designed and deployed voice-first e-commerce assistants using Dialogflow ES, helping increase checkout completions on mobile.
- Built intent-driven workflows with fallback recovery patterns to guide users through transactional flows seamlessly.
- Constructed backend NLP orchestration layers using Node.js, supporting context management and entity resolution.
- Developed NLU training pipelines integrated with form-based user flows, fine-tuned for platform-specific vocabulary.
- Applied semantic text analysis with custom scoring metrics to evaluate input quality and bot understanding.
- Built hybrid mobile modules using React Native, synchronizing user state with dynamic chat UX components.
- Performed weekly audits using Watson NLP metrics, increasing bot accuracy by 28% over six months.
- Championed the introduction of automated regression tests for conversational flows via Vitest.
- Co-authored internal whitepaper on intent mapping architectures and dynamic slot-filling techniques.
- Reduced friction in multilingual transitions by implementing real-time translation middleware.
Software Engineer
Order Group
- Built early NLP integrations in Laravel-based admin tools, enabling voice-guided reporting and status summaries.
- Developed a lightweight text classification pipeline for categorizing system alerts and prioritizing them in dashboards.
- Created JWT-secured endpoints to serve chatbot data from internal systems using RESTful JSON APIs.
- Designed adaptive admin UIs using React.js, configured to respond to voice-triggered intents via bots.
- Integrated chatbot support into WordPress portals, enriching legacy apps with AI-driven assistance.
- Collaborated with PMs to define user stories tied to AI enhancements across B2B portals.
- Delivered auto-reply engines using rule-based inference and fallback ML classification.
- Tracked bot engagement trends across enterprise clients using Cloudflare Analytics and Sentry.
- Conducted voice assistant experiments with command-based shortcuts for enterprise dashboards.
- Co-developed internal semantic tag engine to detect domain-specific language in tickets and conversations.
Web Developer
Fuero Games
- Maintained and enhanced web applications built with PHP, HTML5, CSS3, and JavaScript.
- Created responsive UIs that later served as base structures for chatbot UI experiments.
- Integrated third-party APIs and tracked user events to improve site behavior prediction models.
- Began experimentation with voice UI components embedded in web apps.
- Built CMS extensions for WordPress and Joomla with intent-driven UX workflows.
- Supported bug tracking and hotfix pipelines via GitHub and Jira.
- Developed internal tools for CMS behavior tagging and voice-enabled FAQs.
- Conducted performance profiling using Chrome DevTools and Lighthouse.
- Laid foundational code for secure input handling and token-based sessions.
- Integrated simple REST endpoints to fetch knowledge base content dynamically.
Summary
Senior Software Engineer with over 10 years of experience architecting, training, and deploying production-grade machine learning systems across healthcare, fintech, and conversational AI domains. Adept at bridging the research-to-production gap, I specialize in building deep learning pipelines, transformer-based NLP models, and MLOps workflows that scale across cloud environments. With a strong foundation in statistical learning, real-time model evaluation, and CI/CD automation, I bring a rare blend of curiosity, reproducibility discipline, and systems thinking to ML engineering. Recognized for designing robust, interpretable models and implementing semantic search and agent-based systems aligned with real-world applications.
Skills
Languages
- Python
- Javascript
- Typescript
- Php
- Sql
- Yaml
- Json
- Bash
- Scss
- Html5
- Css3
Ml & Ai Libraries
- Tensorflow
- Pytorch
- Scikit-learn
- Xgboost
- Lightgbm
- Spacy
- Nltk
- Hugging Face Transformers
- Sentence-transformers
- Openai
- Langchain
- Fasttext
- Pandas
- Numpy
- Matplotlib
- Seaborn
- Pydantic
- Mlflow
Llms & Semantic Search
- Gpt-4
- Gpt-3.5
- Rag
- Faiss
- Pinecone
- Weaviate
- Bm25
- Tf-idf
- Elasticsearch
- Bert
- Roberta
- Openai Apis
- Langchain Agents
Frameworks & Tools
- Fastapi
- Flask
- Node.js
- Nestjs
- Express.js
- Laravel
- Django
- Apollo
- Rest
- Graphql
- Websockets
- Trpc
- Next.js
- React
- Vue.js
Cloud & Deployment
- Aws Lambda
- Aws Ecs
- Aws S3
- Aws Cloudwatch
- Aws Fargate
- Azure Functions
- Gcp
- Docker
- Kubernetes
- Jenkins
- Github Actions
- Mlflow
- Airflow
Vector Dbs & Data Infra
- Faiss
- Pinecone
- Chromadb
- Postgresql
- Mysql
- Redis
- Mongodb
- Supabase
- Prisma
- Sequelize
Monitoring & Testing
- Sentry
- Prometheus
- Grafana
- Pytest
- Jest
- Cypress
- React Testing Library
- Vitest
Productivity & Visualization
- Jupyter
- Streamlit
- Dash
- Plotly
- Figma
- Jira
- Slack
- Zeplin
Apis & Sdks
- Openai Sdk
- Google Maps Api
- Stripe Api
- Github Api
- Langchain
- Restful Apis
- Graphql Apis
Methods & Workflow
- Agile
- Scrum
- Kanban
- Tdd
- Dora Metrics
- Rfcs
- Adrs
- Git Flow
Languages
Education
Polish-Japanese Academy of Information Technology
Bachelor's degree, Computer Science · Computer Science · Warsaw, Poland
Similar Freelancers
Discover other experts with similar qualifications and experience