About
Data Scientist focused on building AI systems that work at enterprise scale: reliable, grounded, and safe enough for production.
At SAP I work across the full agentic AI stack. Currently building an agentic workflow platform that combines LLM reasoning, structured retrieval, and external service calls in a deterministic execution engine, with governance and prompt management built in. Before that, a production RAG framework for enterprise search and case management, a prompt framework powering multiple AI services, and auto prompt tuning using grid search, random search, and PPO-based reinforcement learning to drive document-selection accuracy. The load-bearing work across all of it: secure prompt engineering, reasoning chains for explainability, and CI/CD-gated model deployment.
Before SAP, research at DFKI on CV-OCR and NER inside a Trusted Research Environment for sensitive health data, and a Master's thesis at RPTU Kaiserslautern: ASL-LLM, an end-to-end system that generates expressive 3D sign language motion from open-vocabulary text, using a custom dataset of 35k samples built from scratch with YOLOv5 and SMPL-X, achieving 97% token-level accuracy. Earlier, four years at EPAM Systems building a GDPR compliance platform for a global financial institution, covering full-stack microservices, NLP entity extraction from documents, and PCI/PII-compliant database design at production scale.
Experience
-
2024 – present · Germany
Data Scientist · SAP
Agentic workflow platform, enterprise RAG, prompt framework, auto prompt tuning, secure prompt engineering.
-
2023 – 2024 · Germany
Machine Learning Engineer · SAP
Human-in-the-loop feedback service, RAG search evaluation, recommendation quality analysis.
-
2022 – 2023 · Germany
Research Assistant · DFKI / RPTU Kaiserslautern
CV-OCR and NER for secure health record analysis. M.Sc. thesis: ASL-LLM, text-to-3D sign language motion.
-
2018 – 2022 · India
Senior Software Engineer · EPAM Systems
GDPR compliance platform for global financial services: full-stack microservices, NLP entity extraction, PCI/PII compliance.
-
Early career · India
Product Development Intern · Adobe Systems
Performance engineering on the Live Edit Service.
Capabilities
- AI / ML
- LLMs · RAG · Agentic AI · Prompt Engineering · NLP · NER · Computer Vision · OCR · Deep Learning · Reinforcement Learning · Secure ML
- Languages
- Python · Java · JavaScript · Julia · C
- Frameworks
- PyTorch · TensorFlow · LangChain · Hugging Face · spaCy · BERT · FastAPI · Spring Boot · Angular
- Cloud / MLOps
- AWS · GCP · Azure · PCF · Kubernetes · Docker · Kubeflow · CI/CD · Jenkins · MLOps
- Data
- PySpark · PostgreSQL · MongoDB · OpenSearch · Weaviate · Oracle · ArangoDB · Power BI
- Protocols
- MCP · A2A · REST · Microservices