Tip: "Save as PDF" in the print dialog for a clean one-pager.
Gulab Sana Parveen
Data Scientist · Agentic AI, LLMs, RAG, and Enterprise ML Systems
Mannheim, Germany · gulabsanaparveen@gmail.com
Summary
Data Scientist with 7+ years of engineering experience across AI/ML, full-stack systems, and enterprise software. Currently at SAP building agentic AI workflows, production RAG pipelines, and prompt frameworks that power enterprise-scale AI services. Proven track record across the full ML lifecycle: dataset construction, model development, secure deployment, and continuous evaluation. Background spans research (DFKI), 4 years of production software engineering (EPAM), and an M.Sc. in Intelligent Systems from RPTU Kaiserslautern. Focused on systems that are reliable, explainable, and safe for production.
Experience
Data Scientist · SAP · Germany
Oct 2024 – Present- Architecting and contributing to an agentic workflow platform combining LLM reasoning, structured knowledge retrieval, and external service calls in a deterministic execution engine with role-scoped governance, MCP and A2A server interfaces.
- Built and shipped a production prompt framework powering multiple AI services: configurable reasoning depth, three-phase reasoning loop, structured output enforcement, input/output safety moderation, and a reusable task library.
- Designed and implemented a production RAG service routing support queries across multiple documentation types, with LLM-based document selection, two-path prompt routing, PII anonymization, and multi-protocol streaming (SSE/WebSocket).
- Led Auto Prompt Tuning initiative: grid search, random search, PPO-based reinforcement learning, and self-evolving prompt optimization to maximize retrieval accuracy without manual iteration.
- Engineered secure prompt handling to prevent injection attacks and reasoning chains for explainability; deployed model lifecycle on CI/CD with security and code-quality gates.
- Built Claude Code skills for platform onboarding, MCP interface for the execution engine, and translated n8n visual workflows into the typed execution model.
Machine Learning Engineer · SAP · Germany
Oct 2023 – Sep 2024- Designed and implemented a scalable human-in-the-loop feedback service (FastAPI + PostgreSQL) capturing binary, categorical, free-text, and preference signals for continuous AI search quality evaluation.
- Built SHA256-based idempotent session tracking with prompt-edit state archival, transactional validation, and Kubernetes deployment with CI/CD-gated Docker builds.
- Investigated recommendation accuracy degradation across heterogeneous data sources, identifying anonymization inconsistencies and log field noise as root causes; restored parity through targeted pre-processing without model retraining.
- Ran benchmarking and root-cause investigations on ML recommendation systems and search pipelines; led data analysis and dataset creation over large-scale interaction data.
Research Assistant · DFKI / RPTU Kaiserslautern · Germany
Sep 2022 – Sep 2023- Researched CV-OCR (PaddleOCR, EasyOCR, PyTesseract) and NER for a Trusted Research Environment processing sensitive electronic health records (OpenSafely project).
- Contributed to a Secure ML architecture for EHR analysis with privacy-preserving constraints.
- Master's thesis: ASL-LLM, an end-to-end system translating open-vocabulary text into expressive 3D American Sign Language motion. Built a custom 35k-sample dataset using YOLOv5 and SMPL-X; two-stage VQ-VAE and GPT architecture achieving 97% token-level accuracy.
Senior Software Engineer · EPAM Systems · India
Jan 2018 – Feb 2022- Solution architect and lead engineer on a GDPR compliance platform for a global financial institution, delivering from foundations to production over 4 years.
- Designed microservice architecture and PCI/PII-compliant database schema using Spring Boot, PostgreSQL, MongoDB, and Oracle; built Angular frontend for compliance workflows.
- Developed NLP pipeline (Stanford NLP, Apache OpenNLP) for entity extraction from PDFs and images, automating personal data discovery across document repositories.
- Managed CI/CD pipelines with Jenkins and Pivotal Cloud Foundry deployment; worked in Scaled Agile (SAFe) across globally distributed teams.
Product Development Intern · Adobe Systems · India
Early career- Performance engineering on the resource-critical Live Edit Service in Java.
Education
M.Sc. Computer Science · RPTU Kaiserslautern
Apr 2022 – Nov 2024Specialisation: Intelligent Systems, Artificial Intelligence & Data Analytics.
Selected projects
Skills
- Languages
- Python (Pandas, NumPy, PySpark), Java, JavaScript, Julia, C
- AI / ML
- LLMs, RAG, Agentic AI, Prompt Engineering, NLP, NER, Computer Vision, OCR, Deep Learning, Reinforcement Learning, Secure ML
- Frameworks
- PyTorch, TensorFlow, LangChain, Hugging Face Transformers, spaCy, BERT, FastAPI, Spring Boot, Angular
- Cloud / MLOps
- AWS, GCP, Azure, PCF, Kubernetes, Docker, Kubeflow, Jenkins, CI/CD, MLOps
- Data
- PostgreSQL, MongoDB, OpenSearch, Weaviate, Oracle, ArangoDB, Power BI
- Protocols
- MCP, A2A, REST, Microservices