Aniket Kulkarni

About

AI/ML and generative AI developer with experience building end-to-end machine learning pipelines, LLM-based applications, and scalable backend services. Recently completed a Master’s degree in Artificial Intelligence in Germany and is ready for immediate employment.

Experience

• Master's Thesis – Applied AI Engineer
  Robert Bosch GmbH
  Apr 2025 – Sep 2025
• Software Engineer – Enterprise Systems
  NTT DATA
  Sep 2021 – Feb 2023
• Product Development Intern
  Techniken Technologies
  Feb 2021 – Apr 2021
• Project Intern
  Sisai Technologies
  Nov 2020 – Feb 2021

Education

• Master of Science
  University of Applied Sciences Würzburg-Schweinfurt (THWS)
  Artificial Intelligence · Mar 2023 – Mar 2026
• Bachelor of Technology
  Dr. Vishwanath Karad MIT World Peace University
  Electronics and Communication Engineering · Aug 2019 – May 2022

Skills

• SQL
• Computer Vision
• Deep Learning
• FastAPI
• Git
• Python
• Machine Learning
• Prompt Engineering
• MySQL
• NumPy
• Pandas
• Streamlit
• scikit-learn
• OpenCV
• PostgreSQL
• Flask
• PyTorch
• TensorFlow
• AWS
• Docker
• Kubernetes
• LangChain
• Generative AI
• REST APIs
• Natural Language Processing
• Azure
• GCP
• RAG
• LLMs

Projects

• Transformer-based Conversational AI
  BERT, GPT-3, Prompt Engineering, REST API

  Fine-tuned BERT for intent classification with 92.5% accuracy on 220,579 conversation examples and integrated GPT-3 with structured prompt engineering and context-aware memory for agent-based multi-turn response generation. Deployed via a REST API with automated test coverage and evaluation metrics; published at IJRASET 2021.

• Dog Activity Detection and Classification
  ETL, Machine Learning, CNN, Streamlit, REST API

  Built a fully automated ETL and ML pipeline for ingesting raw sensor data, preprocessing, segmentation, CNN model training, validation, and Streamlit web deployment via a REST API. Processed 16,048 real-world data points across 5 classification targets with full test coverage; published at ICDD 2024.

• Airbelts
  Computer Vision, Python

  Built a production AI service for real-time computer vision classification with 82% accuracy; reviewed under the German federal EXIST programme.

• Reinforcement Learning for an Optimal Scheduling Strategy
  Python

  Designed a configurable data and simulation pipeline with 8 agent configurations across 100,000 episodes, including automated metric collection, results logging, and systematic performance evaluation across all variants. Published at IEEE 2025.

Courses & certifications

• MuleSoft Certified Developer Level 1 · Enterprise integration

🗣️ Languages

• German · Intermediate
• English · Fluent
• Hindi · Fluent

📚 Publications

• Optimal Blackjack Strategy with Reinforcement Learning · 2025
• Dog Activity Detection and Classification · 2023
• Evaluation of a Transformer Model for Conversational Chatbots · 2021

🎯 Hobbies & interests

• Research
• AI
• Machine Learning