Industrial IoT: Predictive Maintenance & Financial Lakehouse
A full-stack Medallion architecture pipeline with a ReAct AI Agent, translating raw IoT telemetry into real-time financial risk insights.
Python 3.12
LangChain
Ollama/Llama-3
SQLAlchemy
+1
Hi, I'm
Gitesh Kumar
I build
I specialize in bridging the gap between legacy industrial infrastructure and modern AI. With 4+ years of experience at Hindustan Petroleum, I lead the digital modernization of safety-critical systems, transforming messy SCADA/ERP data into high-performance ML architectures. My expertise lies in Safety-by-Design by building robust RAG frameworks and explainable vision systems for autonomous industrial processes.
Featured Projects
Bringing Academic Research to Industrial Practice
Completed
Explainable AI for End-of-Line Inspection
Industrial defect detection using YOLOv8 and XAI (LRP, Grad-CAM, LIME) to ensure transparency in manufacturing.
Python
PyTorch
YOLOv8
Captum
+2
Enhancing Cybersecurity Intelligence: A RAG Framework for Fine-Tuned LLMs
Master’s Thesis: Developing a synergetic RAG and Fine-Tuning framework to eliminate hallucinations in CVE vulnerability analysis.
Llama-2
PyTorch
FAISS
PEFT/LoRA
Technical Expertise
Tools I use to solve complex industrial problems
AI & Machine Learning
Python
PyTorch
Scikit-learn
Hugging Face
Data & Industrial
SQL
Docker
Git
Linux
Experience
Sr. Operations Officer (Data & Analytics)
Hindustan Petroleum Corporation Limited (HPCL)
Managed end-to-end industrial data lifecycles for a high-capacity LPG plant.
Bridged the gap between legacy SCADA systems and ERP (JD Edwards) to create centralized data models.
Architected automated Python-based ETL pipelines, replacing manual reporting and improving production efficiency by 10%.
Technical Coordinator for field-testing industrial vision systems for safety-critical leak detection.
Education
M.Sc. in Elektromobilität-ACES
Friedrich-Alexander-Universität Erlangen-Nürnberg
Specialization in AI and Sustainable Production Technology.
Master’s Thesis: Developed a Retrieval-Augmented (RAG) framework for fine-tuned LLMs (LLaMA-2) to automate vulnerability analysis in cybersecurity.
Mini Thesis: Implemented XAI techniques (Grad-CAM, LIME) for YOLOv8-based industrial quality inspection at the FAPS Lab.
B.Eng. in Mechanical Engineering
PEC University of Technology, Chandigarh
Get In Touch
Open to Applied AI & Systems Engineering opportunities
Connect
I am always interested in discussing new challenges in Applied AI, data engineering or systems optimization. Whether you are looking to hire, collaborate, or simply exchange ideas, feel free to reach out!
✓ Copied to clipboard!