cv
Basics
| Name | Mihir Agarwal |
| Label | Data Science Graduate Student & Software Engineer |
| ma4874@columbia.edu | |
| Phone | +1 (646) 235-7363 |
| Summary | Software Engineer with experience in Machine Learning and Data Science, currently pursuing a Master's in Data Science at Columbia University. Proven track record at Shell and research labs in building AI-powered automation, predictive maintenance models, and optimizing cloud infrastructure. |
Work
-
2022.08 - 2025.08 Bengaluru, India
Software Engineer
Shell
Engineered AI/ML solutions for HR automation and test case prioritization.
- Built an AI-powered HR Chatbot using a Retrieval-Augmented Generation (RAG) pipeline with LangChain and GPT, indexing Shell HR policy documents for quick, context-aware responses.
- Automated 70% of HR employee inquiries, streamlining internal communications, cutting response times by 50%, and saving $100k annually in operational costs.
- Implemented document chunking, embedding generation, vector indexing, and semantic retrieval for robust intelligent automation systems.
- Developed a web application for Test Case Prioritization using clustering and supervised learning, improving test suite accuracy and efficiency.
- Engineered machine learning models leveraging execution history and step complexity, leading to a 35% reduction in execution time and cost savings of $10k.
-
2022.01 - 2022.08 Paris, France (Remote)
Researcher
LISITE Lab Isep
Research focused on ML-based resource management in fog computing.
- Co-authored and published a Springer journal paper on ML-based resource management in fog computing, proposing latency-reduction strategies.
- Designed and evaluated scalable deployment models by estimating infrastructure requirements and performance metrics, ensuring optimized allocation of computing resources.
-
2020.07 - 2022.08 London, UK (Remote)
Machine Learning Intern
Vidrona
Developed predictive maintenance solutions using computer vision.
- Developed predictive maintenance solutions for insulator caps and guy adjusters by processing drone-captured images with deep learning models (YOLOv3/5/8, Faster R-CNN).
- Enhanced model accuracy through advanced image preprocessing techniques, including Histogram Equalization, noise reduction, and contrast adjustment.
- Optimized detection pipelines by fine-tuning thresholds and anchor boxes, enabling large-scale automation and saving 300+ hours of manual labor weekly.
Volunteer
-
- Present Bengaluru, India
Vice President
Shell Toastmaster Club
Leadership role focused on public speaking and event organization.
- Completed Level 3 and conducted 45+ events to enhance public speaking and leadership skills.
Education
-
2025.08 - 2026.12 New York, NY
Master's
Columbia University
Data Science
- Unsupervised Learning
- Competitive Coding
- Applied Deep Learning
- Probability and Statistics
-
2018.07 - 2022.07 India
Awards
-
Best Domain Prize
IvyHacks
Won Best Domain prize in IvyHacks: a joint hackathon hosted by 6 Ivy League universities.
Certificates
| Deep Learning Nanodegree | ||
| Udacity |
| Neural Networks and Deep Learning | ||
| Coursera |
| Cpp DS and Algorithms, Competitive Coding | ||
| Coding Ninja |
| PCAP Certified Python Programmer | ||
| Python Institute |
Publications
-
ML-based resource management in fog computing
Springer
Proposed latency-reduction strategies that improved fog-cloud integration efficiency.
Skills
| Programming Languages | |
| Python | |
| C/C++ | |
| Java | |
| JavaScript | |
| SQL | |
| HTML/CSS | |
| PHP | |
| XML |
| Frameworks & Libraries | |
| Transformers | |
| TensorFlow | |
| Pytorch | |
| Scikit-Learn | |
| Numpy | |
| Pandas | |
| Matplotlib | |
| OpenCV |
| APIs & Backend | |
| REST API | |
| FastAPI | |
| Flask | |
| Node.js | |
| .NET 8 |
| ML & AI | |
| Machine Learning | |
| Deep Learning | |
| Transformers | |
| BERT | |
| GPT-4 | |
| NLP | |
| RAG | |
| Large Language Models | |
| Computer Vision | |
| Feature Engineering | |
| BLEU | |
| ROUGE | |
| Encoder Models |