Loading AI Portfolio...

Hi, I'm Morris Darren Babu

Structured AI/Data Specialist skilled at transforming and automating data workflows. Proficient in AWS, Azure, and Docker for secure cloud-native DevOps. Expert in GenAI, AI Agents, and LLMOps. Experience at Deutsche Telekom, Volkswagen Group (CARIAD), and BMW Group.

0
% Security Detection Boost
0
% Diagnostic Precision
0
% ETL Pipeline Speed
Explore Research Download CV Get In Touch LinkedIn
Deep Learning
Automotive AI
Machine Learning
Scroll to explore

About Me

Structured AI/Data Specialist driving innovation through cutting-edge AI and data engineering

AI/Data Specialist

I'm currently working as an MLOps Engineer Intern at Deutsche Telekom, building LLM benchmarking frameworks and scaling automated pipelines with Kubernetes. Previously, I completed my Master's thesis at Volkswagen Group (CARIAD), integrating Large Language Models in automotive CI/CD/CT pipelines for AI-enhanced security testing.

With experience at BMW Group in RAG & LLM systems, and expertise in GenAI, AI Agents, LLMOps, computer vision, and cloud-native DevOps, I transform complex technical challenges into innovative solutions. Proficient in AWS, Azure, Docker, and Kubernetes for secure deployments.

M.Sc. Data Science

Friedrich Alexander University Erlangen-Nuremberg (2022-2026)

Location

Bonn, NRW, Germany

Email

morrisdarren357@gmail.com

Technical Expertise

AI & Machine Learning

TensorFlow
PyTorch
Scikit-learn
Computer Vision
NLP / NLTK
PyTorch Geometric
Tesseract OCR

GenAI & LLMOps

Prompt Engineering
RAG
GAN
VAEs
LangChain
vLLM
MLflow
ClearML
Hugging Face
LoRA / QLoRA

Data Science & Analytics

Pandas
NumPy
Matplotlib
Seaborn
Power BI
Snowflake
Apache Spark
ETL Pipelines

Cloud & DevOps

AWS (EC2, Aurora, SageMaker, S3, CloudWatch)
Azure (Spark, Data Factory, Data Lake)
Docker
Kubernetes (Helm)
GitLab CI/CD
GitHub Actions
Git
Keycloak
n8n
Grafana

Programming & Frameworks

Python
R
C++
JavaScript
SQL
VBA
Flask
Django
Node.js

Databases

MySQL
PostgreSQL
MongoDB
Cassandra
Snowflake
SAP R3

Current Research

AI Usage in CI/CD/CT Pipelines for Automotive Compute Platforms

Master Thesis Research

Volkswagen Group (CARIAD)

May 2025 - September 2025

Research on integrating LLMs in automotive systems, implementing AI-driven black-box fuzzing in Azure AI Foundry by benchmarking 16 LLM models to strengthen ECU resilience and reduce OEM cybersecurity threats.

Key Research Objectives

  • AI-Driven Black-Box Fuzzing with LLMs
  • Containerized LLM Inference in CI/CD/CT Pipelines
  • Automated Test-Case Generation
  • Secure Azure Private Link for Fuzz-Testing PoC
  • Benchmarking 16 LLM Models for Automotive Security

Research Impact

+7%
Code Coverage Boost
+13%
Code-Flaw Detection
-33%
Test Creation Time
-10%
Execution Delays

Professional Experience

Journey through leading automotive and technology companies

MLOps Engineer Intern | LLM Infrastructure & Deployment

Deutsche Telekom

Bonn, Germany

October 2025 - March 2026

  • Built a microservices-based LLM benchmarking framework on vLLM using GuideLLM (latency, throughput) and DeepEval (LLM-as-judge), tracking experiment runs in ClearML to define SLOs for model selection and capacity planning
  • Scaled automated benchmarking via GitLab CI/CD and Kubernetes runners, executing containerized matrix jobs on Hugging Face datasets, publishing reports to Grafana
  • Secured enterprise MLflow deployments on Kubernetes, implementing Keycloak OIDC and OAuth sidecars for SSO, enforcing ingress, TLS, and RBAC to meet compliance
vLLM Kubernetes (Helm) GitLab CI/CD MLflow ClearML Keycloak Grafana Docker Hugging Face

Master Thesis | Integrating LLMs In Automotive Systems

Volkswagen Group (CARIAD)

Wolfsburg, Germany

May 2025 - September 2025

  • Implemented AI-driven black-box fuzzing in Azure AI Foundry by benchmarking 16 LLM models, boosting code-flaw detection by 13%, strengthening ECU resilience, and reducing OEM cybersecurity threats
  • Automated test-case generation by containerizing LLM inference services in CI/CD/CT pipelines, cutting test creation time by 33% and increasing code coverage by 7%
  • Designed a secure Azure Private Link for an automated LLM fuzz-testing PoC, replacing manual tests, reducing delays by 10%, and improving automotive software quality
LLMs Azure AI Foundry CI/CD/CT Black-box Fuzzing Python Docker Automotive AI

Data Scientist Intern | RAG & LLM Systems

BMW Group

Munich, Germany

October 2024 - April 2025

  • Deployed a CRNN surface detection model on AWS SageMaker (EC2, S3) with semantic segmentation, tuning AI parameters and serving results via API to cut false positives
  • Architected a Django Web application for parsing geometric STL meshes and file ingestion, utilizing document chunking pipelines to achieve 8% faster processing
  • Engineered an LLM auto suggestion model utilizing RAG and vector search to identify part issues, via document chunking and data indexing to elevate precision by 12%
AWS SageMaker RAG LLMs Django Computer Vision Python

Data Analyst

Crystal Consultancy Services

Chennai, India

October 2021 - September 2022

  • Monitored and analyzed key performance indicators (KPIs) using SAP R3 data, resulting in a 4% improvement through targeted data-driven decisions
  • Applied Azure Databricks to address slow data processing in the ETL pipeline, optimizing feature engineering and model deployment, enhancing speed by 25%
SAP R3 Azure Databricks ETL Power BI SQL

Data Engineering Intern

Altascio Technologies

Texas, USA (Remote)

February 2021 - July 2021

  • Utilized Excel techniques (Pivot, VBA, VLOOKUP, SUMIFS) to streamline manual aggregation of department spreadsheets, reducing report generation time by 2%
  • Streamlined data extraction and consolidation using SQL to resolve fragmented inventory data, saving 20 hrs/week and reducing inventory update time by 11%
Excel VBA SQL Data Consolidation VLOOKUP Pivot Tables

Python Developer Intern

Jman Group

London, United Kingdom (Remote)

June 2020 - December 2020

  • Assisted in developing predictive models with ML algorithms using sensor data to identify clusters, detect outliers and trends, delivering 87% accuracy in field tests
  • Applied end-to-end data science (data collection, EDA, feature engineering, model selection, deployment) to solve ETL pipeline bottlenecks, raising throughput by 27%
Python Machine Learning ETL Pipelines Data Science Scikit-learn

Featured Projects

Innovative solutions in AI, ML, and automotive technology

Graph Neural Networks

Material Wear Prediction with GNNs

Constructed a custom GNN framework and synthetic dataset of shape/material data, predicting wear with RMSE under 0.1 and supporting end-to-end shape generation and graph visualization.

PyTorch PyTorch Geometric GATv2 GraphSAGE Scikit-learn
RMSE < 0.001 Test Accuracy
Computer Vision & NLP

OCR + Auto Spelling Correction

Implemented an NLP-driven spelling correction system using tokenization, language modeling, and Tesseract OCR, exposing it via a scalable REST API and reducing Character Error Rate by 89% on multilingual documents.

Tesseract OCR NLP Flask Python REST API
89% Character Error Rate Reduction
Machine Learning

Hybrid Movie Recommendation System

Created a hybrid recommendation system using collaborative and content filtering with clustering, achieving a 92% Top-K Hit Rate and integrating it into a Flask-based web app for real-time user recommendations.

Collaborative Filtering Content Filtering Flask Clustering
92% Top-K Hit Rate
Security & LLMs

LLM Fuzz Monitor

AI-driven fuzz-testing toolkit that benchmarks LLM-generated fuzz drivers for C/C++ projects, detects hallucinations, analyzes code quality & vulnerabilities, and orchestrates large-scale comparative experiments across 14+ models and 60+ repositories.

Python Ollama OpenAI Docker SQLite Rich CLI
14+ LLM Models Benchmarked
60+ C/C++ Repos Tested
SaaS & Privacy

ConsentHub — B2B Consent Platform

Production-ready B2B SaaS platform for managing data-subject consent across GDPR, CCPA, and LGPD frameworks. Features JWT-based RBAC, multi-tenancy, analytics dashboards, CSV/JSON export, and a Kubernetes Helm deployment with CI/CD.

Flask SQLAlchemy JWT Auth Docker Helm / K8s GitHub Actions
24+ Automated Tests
3 Privacy Frameworks
AI & Marketing Automation

Relevance Engine — AI Next-Best-Action

Consent-first, AI-powered marketing automation platform with a next-best-action decision engine. Features behavioral scoring, fatigue management, journey orchestration, A/B experimentation, and pluggable AI agents — built on FastAPI, PostgreSQL, and Redis.

FastAPI PostgreSQL Redis SQLAlchemy Docker Helm / K8s
20+ REST API Endpoints
5 AI Agent Modules
Workflow Automation

IntentGraph — Action OS

A universal action OS that compiles natural-language intent into trusted, auditable workflows. Features a typed plugin contract with preview/execute/compensate phases, policy-driven approval gating, and full rollback support — deployed via Terraform on AWS with Helm charts.

TypeScript Node.js Terraform Helm / K8s GitHub Actions
58 Tests Across 5 Suites
8 Policy Rules

Publications & Research

Contributing to academic knowledge and industry innovation

Integrating LLMs in Automotive CI/CD/CT Pipelines for AI-Enhanced Security Testing

Morris Darren Babu — Master Thesis, Volkswagen Group (CARIAD) & Friedrich Alexander University

Investigated the integration of Large Language Models into automotive CI/CD/CT pipelines for black-box fuzz testing. Benchmarked 16 LLM models in Azure AI Foundry, automated test-case generation via containerized inference services, and designed a secure Azure Private Link architecture — boosting code-flaw detection by 13%, cutting test creation time by 33%, and increasing code coverage by 7%.

16 LLM Models Benchmarked
+13% Code-Flaw Detection
2025 Published
View Research Details

Movie Recommendation System using Hybrid Filtering

Morris Darren Babu

Created a hybrid recommendation system using collaborative and content filtering with clustering, achieving a 92% Top-K Hit Rate and integrating it into a Flask-based web app for real-time user recommendations.

92% Top-K Hit Rate
2022 Published
View on GitHub

Research Accomplishments

Predictive Data Analytics Nanodegree

Udacity — Bertelsmann Scholarship

Devised 25+ analytical datasets during 10-month program, utilizing statistical modeling, classification, time-series forecasting, and A/B testing to improve analytical efficiency by 17%.

Germany GDP Prediction Research

OSS Research Group

Forecasted Germany's GDP from 1970 to 2022 in a 6-month open source project by refining time-series data, analyzing inflation to GDP correlation and evaluating models via RMSE and MAE metrics.

Material Wear Prediction with Graph Neural Networks

University Research Project

Constructed a custom GNN framework and synthetic dataset of shape/material data, achieving RMSE under 0.001 in wear prediction using GATv2 and GraphSAGE architectures with end-to-end shape generation and graph visualization.

Certifications & Education

Continuous learning and professional development

M.Sc. in Data Science

Friedrich Alexander University Erlangen-Nuremberg

October 2022 - March 2026

Pattern Recognition
Deep Learning
Generative AI
Anomaly Detection
Reinforcement Learning
Decision Theory
Statistics
ML in Finance

B.E. in Computer Science Engineering

Rajalakshmi Institute of Technology, India

June 2018 - June 2022

Data Structures
Machine Learning
Database Systems
Computer Networks
Software Engineering

Google Cloud Platform (GCP)

IT Automation with Python Specialization

Deep Learning AI

TensorFlow Developer Specialization

Data Warehousing & Big Data on GCP

Cloud Data Engineering

Excel to MySQL

Regression & Statistics Techniques

Predictive Data Analytics Nanodegree

Udacity – Bertelsmann Scholarship

Languages

English C1 (Business Fluent)
German A2 acquired, B1 ongoing
French A1
Tamil Native

Let's Connect

Ready to collaborate on innovative AI and automotive projects

Email

morrisdarren357@gmail.com

Send Email

Phone

+49 157 81921462

Call Now

Location

Bonn, NRW, Germany