Devakinand D.

Master's Thesis: Analyzing Prompt Engineering for Data Extraction from Unstructured Data

Rosenheim, Germany

Experience

Nov 2024 - May 2025
7 months

Master's Thesis: Analyzing Prompt Engineering for Data Extraction from Unstructured Data

Technical Institute of Rosenheim

  • Applied advanced machine learning techniques by developing a multi-strategy prompting framework (zero-shot, few-shot, CoT, instruction tuning) to extract structured data from complex financial and medical datasets, significantly enhancing model reliability and achieving an 18% improvement in F1-score through evaluation using ROUGE-L, METEOR, and Cosine Similarity.
  • Designed scalable structured-output workflows and built automated monitoring pipelines (spaCy, ClearML) for continuous performance tracking, simulating real-world MLOps principles.
  • Refined prompt strategies iteratively based on meticulous error analysis to ensure robust, production-ready performance.
Mar 2024 - Aug 2024
6 months

Student Assistant - AI Validation & Prompt Engineering

Technical Institute of Rosenheim

  • Designed and deployed LLM pipelines (GPT-3.5, GPT-4) for document-to-structured data conversion, improving automation and data accuracy.
  • Engineered end-to-end MLOps pipelines for continuous evaluation and reliable LLM deployment, leveraging ClearML to orchestrate CI/CD-like workflows.
  • Conducted bias detection and mitigation strategies, improving cross-domain robustness and fairness by 20%.
  • Collaborated with IT and research teams to integrate AI agents into existing systems and client environments.
Sep 2023 - Feb 2024
6 months

Student Assistant - Explainable AI for Real-Time Prediction

Technical Institute of Rosenheim

  • Designed and deployed an Explainable AI (XAI) framework (SHAP and LIME) for predictive modeling in manufacturing (injection molding), improving transparency and driving higher adoption rates of AI solutions.
  • Automated data preprocessing to accelerate model readiness and ensure rapid deployment and iteration in production environments.
  • Collaborated with domain experts to validate model results, ensuring alignment with real-world operational needs.
  • Developed advanced visualization dashboards to communicate AI insights to non-technical, cross-functional teams.
Oct 2021 - Jul 2022
10 months

Synthetic Data Generation for Optical Localization (Master's Project)

Technical Institute of Rosenheim

  • Generated over 50,000 labeled synthetic images using Blender's 3D rendering capabilities for object detection model training, significantly enhancing algorithm generalization.
  • Created automated annotation scripts to generate bounding boxes, rotation angles, and pixel coordinates for every frame, ensuring a consistent and scalable labeling process.
  • Integrated synthetic data into training pipelines for YOLO and Faster R-CNN models, achieving a 15% improvement in detection accuracy on real-world test datasets.

Summary

Master's graduate and hands-on Al/ML Scientist specializing in designing and deploying production-focused Deep Learning, LLM-powered systems, and scalable data & Al solutions. Proven ability to translate complex data science prototypes into functional tools that drive measurable business impact. Eager to apply strong analytical and technical skills in a client-facing, consultancy environment.

Core Expertise: Cutting-edge Machine Learning Techniques, MLOps (ClearML), Prompt Engineering, Explainable AI (XAI), and Cross-functional Collaboration to deliver impactful, trustworthy, and algorithmic solutions.

Languages

Telugu
Native
German
Advanced
English
Advanced

Education

Oct 2020 - Jul 2025

Technical Institute of Rosenheim

Masters in Engineering Sciences, Mechatronics · Rosenheim, Germany

Aug 2015 - May 2019

SRM Institute of Science and Technology

Bachelor of Technology, Mechanical Engineering · Chennai, India

Certifications & licenses

Convolutional Neural Networks With TensorFlow

Coursera

Crash Course On Python

Coursera

Machine Learning Specialization

Coursera

Neural Networks And Deep Learning

Coursera

Need a freelancer? Find your match in seconds.
Try FRATCH GPT
More actions