Bank annual-report intelligence
Building a retrieval system for European bank reports with structured tables, page-level provenance and grounded financial analysis.
In progress · 2026DATA SCIENTIST / AI ENGINEER
Making AI useful.
Making systems last.
I’m Prabhupad Pradhan, a data scientist at Deutsche Bank. I work across applied AI, information retrieval and cloud engineering—turning ambitious ideas into dependable systems.
● Frankfurt am Main, Germany
Deutsche BankApplied AI · Financial services
ECML PKDD ’24Published NLP research
DAAD PrizeAcademic recognition · 2023
M.Sc. · 1.4Applied Data Science
01 / SELECTED WORK
Ideas meet execution.
Research, engineering projects and work in progress—built with rigor and real-world impact.
Pointer-guided pre-training
Co-authored research on paragraph-level context and segment ordering for language models, published at ECML PKDD 2024.
Explore ↗ESRS disclosure retrieval
Structured document retrieval for sustainability reporting, compliance concepts and traceable evidence.
Explore ↗Smart traffic light system
YOLO-based object detection and priority-aware traffic management, including emergency-vehicle recognition.
Explore ↗Real-time sentiment platform
A GKE-hosted sentiment analysis system with inference microservices, CI/CD and a monitoring dashboard.
Engineering projectPersonal finance intelligence
Prototyping expense classification, transaction analytics and multi-agent financial insights.
In progress · 202602 / EXPERIENCE
Research meets real impact.
From enterprise software to applied research and production AI in financial services.
Full professional profile ↗Data Scientist · Deutsche Bank
Developing fraud-prevention data science solutions in Private Bank, including Python and Node.js services, real-time features, Google Cloud Run deployments and CI/CD.
Data Engineer · Mediaplus Group
Worked in data engineering, connecting earlier enterprise automation experience with data-oriented software and analytics workflows.
Research Assistant · Fraunhofer IAIS
Research in NLP and information retrieval for financial and legal documents, including large-scale text processing and model evaluation.
Python & Automation Engineer · Accenture
Built Python automation, testing frameworks and CI/CD tooling for enterprise software reliability.
Software Engineer · Amdocs
Telecommunications software systems, automated testing and complex release and migration work.
03 / RESEARCH & RECOGNITION
Curious by nature.
Rigorous by design.
Research informs how I build: question assumptions, evaluate carefully and make outcomes reproducible.
ECML PKDD 2024 · PEER-REVIEWED
Pointer-Guided Pre-Training: Infusing Large Language Models with Paragraph-Level Contextual Awareness
Research on paragraph-level coherence, segment ordering and language-model representations.
Read on arXiv ↗ACADEMIC RECOGNITION
DAAD Prize2023 · Academic achievement
ACADEMIC DISTINCTION
Best Student of the YearSRH Hochschule Heidelberg · November 2024
PROFESSIONAL RECOGNITION
SPOT Award for InnovationFebruary 2021 · Recognized for an innovative ETL automation suite.
EDUCATION
M.Sc. Applied Data ScienceSRH Hochschule Heidelberg · 2022–2024 · Final grade 1.4
MASTER’S THESIS · OCT 2023 — MAR 2024
Exploring the Capabilities of LLMs in Information Retrieval SystemsDeveloped a retrieval-augmented generation pipeline that matches reporting requirements with relevant passages. Tools included Python, LangChain, LangSmith, Hugging Face, PyTorch Lightning, and Weights & Biases.
CERTIFICATIONS
Continuous technical learningNVIDIA Fundamentals of Deep Learning (2023) · SAS Visual Analytics 2 for SAS Viya: Advanced (2022) · Machine Learning, Stanford University / Coursera (2021).
04 / ABOUT & EXPERTISE
Thoughtful research.
Practical engineering.
My background spans enterprise software engineering, NLP research and production AI. I focus on trustworthy AI, document intelligence, information retrieval and the infrastructure that makes machine learning useful beyond a demo.
01 / Applied AI & Research
LLMs & RAGNLPInformation RetrievalPyTorchHugging Face
02 / Software & Cloud
PythonFastAPIGoogle CloudDockerKubernetesGitHub Actions
03 / Data & MLOps
SQLPostgreSQLMongoDBPySparkDatabricksMLflow
05 / LET’S CONNECT
Good ideas deserve
to be built well.
Interested in AI engineering, document intelligence or a thoughtful technical conversation?