Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

Hi, I'm Nikhil Mankali, a Machine Learning Engineer focused on building production-ready vision models and large language models for real-world applications. I design and deploy AI systems using PyTorch, OpenCV, Docker, Kubernetes, and AWS to solve complex problems and deliver measurable impact. My specialties include semantic segmentation, object detection, OOD detection, and adversarial robustness, with experience turning research ideas into robust CV pipelines and reliable real-time solutions. I enjoy collaborating across teams and advancing uncertainty estimation to improve model reliability and decision-making.…Hi, I'm Nikhil Mankali, a Machine Learning Engineer focused on building production-ready vision models and large language models for real-world applications. I design and deploy AI systems using PyTorch, OpenCV, Docker, Kubernetes, and AWS to solve complex problems and deliver measurable impact. My specialties include semantic segmentation, object detection, OOD detection, and adversarial robustness, with experience turning research ideas into robust CV pipelines and reliable real-time solutions. I enjoy collaborating across teams and advancing uncertainty estimation to improve model reliability and decision-making.

Nikhil Mankali

AI Engineer, Data Scientist, DevOps Developer, +4





Hi, I'm Nikhil Mankali, a Machine Learning Engineer focused on building production-ready vision models and large language models for real-world applications. I design and deploy AI systems using PyTorch, OpenCV, Docker, Kubernetes, and AWS to solve complex problems and deliver measurable impact. My specialties include semantic segmentation, object detection, OOD detection, and adversarial robustness, with experience turning research ideas into robust CV pipelines and reliable real-time solutions. I enjoy collaborating across teams and advancing uncertainty estimation to improve model reliability and decision-making.…Hi, I'm Nikhil Mankali, a Machine Learning Engineer focused on building production-ready vision models and large language models for real-world applications. I design and deploy AI systems using PyTorch, OpenCV, Docker, Kubernetes, and AWS to solve complex problems and deliver measurable impact. My specialties include semantic segmentation, object detection, OOD detection, and adversarial robustness, with experience turning research ideas into robust CV pipelines and reliable real-time solutions. I enjoy collaborating across teams and advancing uncertainty estimation to improve model reliability and decision-making.

Available to hire

Hi, I’m Nikhil Mankali, a Machine Learning Engineer focused on building production-ready vision models and large language models for real-world applications. I design and deploy AI systems using PyTorch, OpenCV, Docker, Kubernetes, and AWS to solve complex problems and deliver measurable impact.

My specialties include semantic segmentation, object detection, OOD detection, and adversarial robustness, with experience turning research ideas into robust CV pipelines and reliable real-time solutions. I enjoy collaborating across teams and advancing uncertainty estimation to improve model reliability and decision-making.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Language

German

Fluent

English

Fluent

Work Experience

Research Assistant at Fraunhofer IVI AWZ Ingolstadt

November 1, 2023 - Present

Designed, developed, and deployed a pixel-wise OD detection module for semantic segmentation in aerial perception systems; computed performance metrics (AUROC) and validated in real-world drone flight tests; integrated uncertainty estimation into CV pipeline to improve reliability of object recognition and decision-making. Used Docker, Kubernetes, Kubeflow, Grafana, and Git to build, monitor, and maintain reproducible, deployment-ready CV pipelines. Collaborated with internal teams on model testing, containerized vision modules, and cloud-based experimentation.

Student Assistant at Technische Hochschule Ingolstadt

June 1, 2023 - Present

Gathered, extracted, and cleaned datasets with various view types (top-down, frontal, oblique) to create a unified aerial vision dataset; balanced dataset coverage and ensured annotation quality for training object detection and segmentation models. Resulted in improved accuracy and generalization of downstream computer vision tasks.

Research Study Assistant at University of Siegen

August 1, 2022 - Present

Investigated adversarial robustness of CV models by implementing various attacks on grayscale, RGB, and 3D voxel data; calculated and benchmarked model robustness metrics; improved adversarial resistance using adversarial training; results contributed to a PhD thesis on vision model safety in adversarial environments.

Research Assistant at Fraunhofer IVI AWZ, Ingolstadt, Germany

November 1, 2023 - November 27, 2025

Designed and deployed a pixel-wise OOD detection module for semantic segmentation in aerial perception systems, achieving AUROC; validated in real-world drone flight tests; integrated uncertainty estimation into the CV pipeline to improve reliability of object recognition and decision-making. Used PyTorch, OpenCV, Docker, Kubernetes deployment via kubectl, and AWS; collaborated with internal teams to support model testing, containerizing modules, and cloud-based experimentation.

Intern at Fraunhofer IVI AWZ, Ingolstadt, Germany

June 1, 2023 - November 27, 2025

Analyzed and harmonized 15 input normalization; wrote scripts to automate label conversion and enhance compatibility with object detection and segmentation frameworks. Created internal documentation on Docker image creation, Kubernetes deployment via kubectl, and Octane simulation usage; now used as onboarding material.