Nadeem Nazer | AI Researcher & Developer

Who I Am

About Me

I'm an AI Research Engineer and Software Developer holding Master's with specialisation in Deep Learning & Generative Models from Otto von Guericke University, Magdeburg. With over 4+ years of industry experience, I have worked on projects spanning computer vision, NLP, and full-stack web development.

I'm passionate about pushing the boundaries of AI - from fine-tuning large vision-language models to building scalable web applications. I thrive where research rigour meets practical engineering.

4+ Years Experience

8+ Projects

4 Companies

LinkedIn GitHub

Magdeburg, Germany

M.Sc. Digital Engineering (OVGU)

AI Research Intern @ BCAI

Python · PyTorch · React · FastAPI

English (Fluent) · German (B1)

Career

Experience

AI Research Intern Full-time

Bosch Center for Artificial Intelligence (BCAI), Germany

Jan 2025 – Present

Researched and developed SOTA Deep Learning models for scene understanding using RT-DETR and YOLOP to deconstruct raw video data into semantic JSON representations, capturing object trajectories.
Implemented perspective transformation algorithms to map 2D pixel-level detections into 3D metric coordinates, ensuring spatially accurate inputs for the downstream generation module.
Developed an Agentic framework that consumes the JSON data to generate novel, adversarial driving scenarios.
Leveraged Azure OpenAI GPT-4o model to parametrically augment scenes, applying RAG to ensure generated scenarios adhere to traffic regulations.
Deployed the entire pipeline as microservices, using FastAPI , and Next.js, containerized via Docker and automated for production using GitHub Actions CI/ CD.
Researched and developed a novel prompt-tuning framework for Vision-Language foundation models (CLIP) on Anomaly detection domain, achieving SOTA results on public datasets (avg. +4% accuracy) and a 35% performance increase on internal proprietary datasets
Built a MLOps pipeline using MLFlow logging experiments, hyperparameters and managing model artefacts during ablation studies.
Deployed the final model using Docker and FastAPI allowing zero-shot inference for real-time visual inspection.

PyTorchAgentic AILVLMsComputer VisionDeep Learning

Research Assistant – Full Stack Developer Part-time

Otto Von Guericke University, Germany

Aug 2024 – Dec 2024

Developed a Health-LLM system by fine-tuning BioMistral-7B model on tele-clinic data.
Implemented end-to-end RAG pipeline to retrieve related patient health reports from vector databases , and engineering few-shot examples to restrict LLM from hallucinating and generate context relevant response.
Contributed to open-source project involving discrete mathematics proofs and logical reasoning (fitch.js)

LLMsRAGJavaScriptIn-context learning

Research Assistant – Python Developer Part-time

Fraunhofer IAIS, Germany

Oct 2023 – Apr 2024

Developed REST API endpoints using FastAPI.
Wrote comprehensive unit and integration tests with Pytest, including fixtures and mocks.
Refactored legacy code with dependency injection, improving performance and maintainability.

FastAPIPythonPytestREST API

Research Assistant – Full Stack Developer Part-time

Otto Von Guericke University, Germany

Aug 2022 – Sep 2023

Designed and developed a responsive web application from scratch using React and Flask.
Migrated the codebase to Next.js for improved performance and SEO.
Integrated Prisma ORM to optimise database operations.

ReactNext.jsFlaskPrisma

Application Developer Full-time

Tata Consultancy Services, India

Jul 2019 – Mar 2022

Developed web components for Bayer's global website using Drupal, SCSS, and Twig in an Agile team.
Worked on PHP (Drupal 8) backend solutions for data management and content workflows.
Implemented client-requested enhancements that improved user experience metrics.
Mentored junior developers and led resolution of critical bugs.

Drupal 8PHPSCSSAgile

Tech Stack

Skills

AI / ML

PyTorch

TensorFlow

Scikit-learn

NumPy

Pandas

OpenCV

Jupyter

Languages

Python

JavaScript

TypeScript

HTML5

CSS3

Frameworks

React

Next.js

Flask

FastAPI

Tailwind

Streamlit

Material UI

Databases & Cloud

MySQL

MongoDB

GCP

SQLAlchemy

Tools & DevOps

GitHub

GitLab

CI/CD

Pytest

Linux

Docker

npm

What I've Built

Projects

Unsupervised Image Classification

Grouped images into semantically meaningful clusters without ground truth using the SCAN method. Applied Grad-CAM to highlight important image features.

PyTorchPythonComputer Vision

OMDB Browser

Movie search app with metadata display, caching, and recommendation functionality. Built with Next.js and Tailwind CSS.

Next.jsTailwind CSSAPI

Health-LLM: RAG Disease Diagnosis

Personalised health diagnosis using BioMistral-7B and LangChain with RAG techniques, vector databases, and few-shot examples for context-relevant responses.

PythonLangChainRAGLLM

MLOps Real-Time Object Detection

Real-time image segmentation app with Streamlit and YOLOv8, deployed on AWS SageMaker with a full MLOps pipeline for continuous monitoring.

StreamlitYOLOv8AWS SageMaker

Interview Automation System

Intelligent QA system using Word Mover's Distance and a Google News word vector model to assess semantic similarity between candidate and expected answers.

PythonFlaskNLPOpenCV

Time Series Forecasting – VertexAI

Built and deployed a forecasting model on Google VertexAI. Visualised results and created an interactive dashboard with Google Looker Studio.

GCPVertexAILooker Studio

Chatbot for E-Commerce

AI chatbot handling common e-commerce queries built using a feed-forward Neural Network with 2 hidden layers, served via a Flask backend.

PyTorchFlaskJavaScript

Search Result Clustering

Web app to search a topic and visualise clusters of Wikipedia documents using Elasticsearch and K-Means clustering with an interactive React frontend.

ReactPythonElasticsearchK-Means

Academic Background