Hi, I’m Maya Devarajan.

MBSE Software Architect • Seattle, WA

I’m passionate about using data and machine learning to solve meaningful challenges and drive smarter decisions. I love working with curious, forward-thinking teams where technical depth meets creative problem-solving.

View Projects Email Me Resume

About

I’m driven by the challenge of using data and machine learning to create meaningful impact in complex, real-world settings. I’m especially motivated by work that sits at the intersection of data, technology, and collaboration — where analytical rigor meets creative problem-solving. I’m interested in opportunities in data science, ML engineering, or analytics that allow me to apply technical depth, foster innovation, and contribute to teams shaping the future of intelligent systems.

Experience

MBSE Software Architect · Dassault Systèmes

2024–Current · Seattle, WA

Skills
- Systems Modeling
  - SysML and UAF for complex, distributed architectures
  - Reliability assessment across structural and behavioral views
- Data & ETL
  - ETL pipelines to extract, transform, and load engineering models
  - Feature engineering from architecture artifacts for analytics
- ML Integration
  - Interface predictive ML models with architecture frameworks
  - Automated ingestion and real-time decision dashboards
- Analytics Methods
  - Uncertainty quantification and sensitivity analysis for KPIs
  - Data-informed decisions under system variability
- Tools & Languages
  - Python, SQL; PyTorch/TensorFlow; Web services
Vaccine Manufacture Data Analytics Intern · Merck & Co., Inc.

Jun 2024–Aug 2024 · West Point, PA

Skills
- Analytics & BI
  - Power BI, Power Query, DAX, SQL for quality/compliance ops
  - Dashboards for disinfection monitoring and sample status
- Solutions Delivered
  - Three production-ready solutions to improve efficiency
  - Quality notification and manufacturing workflow improvements
- Data Engineering
  - Automated transformations and pipeline optimization
  - Improved dashboard responsiveness and maintainability
Machine Learning Researcher — Cornell Tech

Jan 2022 — Aug 2023

Skills
- NLP & Privacy
  - BERT-based de-identification for electronic health records (EHRs)
  - Natural Language Processing for medical text
- Modeling & Evaluation
  - Model monitoring with black-box pipelines in PyTorch
  - Cross-validation, metrics, and statistical analysis
- ML Stack
  - Python, PyTorch; model fine-tuning and regularization

Projects

pdf-to-rag

Local-first RAG pipeline over PDF folders: PDF → text normalization → chunked passages with metadata → embeddings (Transformers.js / ONNX by default, optional Ollama) → versioned JSON/binary index → semantic query returning ranked verbatim excerpts with file name and page for citation. CLI, library, and MCP (stdio or HTTP) share the same application layer (ingest, query, inspect); GitHub Actions CI and layered architecture (commands, application, domain, pipeline modules).

Node.js • TypeScript • Transformers.js • RAG • MCP • CI

Code

Contact

Interested in working together? Reach out via email or connect on LinkedIn.