Production LLM Systems
Built schema-constrained generation pipelines with Pydantic validation, retry logic, and auto-repair at NEC Labs — not toy demos.
Hi, I'm Amey Narwadkar
Working Student at NEC Labs Europe · M.Sc. Scientific Computing, Heidelberg
About
I work at the intersection of LLM systems, agentic workflows, and RAG pipelines, with a strong mathematical background and hands‑on experience building reliable, observable AI systems.
- Working Student (Software Development) at NEC Labs Europe, building LLM-powered prototypes, structured generation pipelines, and agent evaluation tools.
- M.Sc. Scientific Computing student at Universität Heidelberg (AI, NLP, efficient methods, numerical methods).
- Enjoy building end‑to‑end projects – from LLM pipelines and agent benchmarks to clean UX and deployment.
Experience
NEC Labs Europe
Working Student — Software Development- Developed internal LLM-powered prototypes for document intelligence, structured generation, and retrieval-augmented workflows.
- Built multi-source ingestion pipelines (PDFs, URLs, text) with chunking, retrieval, source attribution, and structured output generation.
- Designed schema-constrained LLM pipelines with JSON/Pydantic validation, retry logic, and automatic repair for malformed outputs.
- Contributed to synthetic longitudinal data-generation workflows with cross-artifact consistency, RAG-backed reports, and token/cost tracking.
Soul AI
AI Data Specialist — LLM Evaluation- Contributed to RLHF data pipelines evaluating LLM outputs for college-level mathematics — multi-turn reasoning, hallucination reduction, and context-retention analysis.
- Reviewed and annotated model outputs across 500+ evaluation tasks, identifying failure patterns and collaborating with human evaluators to improve model quality.
Education
Universität Heidelberg, Germany
Master's in Scientific ComputingFocusing on Advanced methods in AI, efficient methods in NLP, Advanced applied mathematics, and Computer vision.
Fergusson College, Pune
Bachelor's in MathematicsBuilt a strong foundation in mathematical concepts, statistical analysis, and problem-solving techniques for AI/ML.
Tech Stack
What I bring
Agent Architecture Research
Benchmarked 11 LLM agent architectures (ReAct, Plan-Execute, Reflection, Debate, MoA) with real-time trace analysis and failure diagnosis.
Math → ML Foundation
B.Sc. Mathematics + M.Sc. Scientific Computing at Heidelberg. Comfortable going from paper to implementation in PyTorch.
Proof of Work
My Project Journey
A chronological showcase of projects I've built, from foundational learning to cutting-edge research.
$
Contact
Let's get in touch!
I’m happy to chat about internships, research collaborations, or full‑time ML roles. Best way to reach me is via email or LinkedIn.