Kartikey Kawadkar
Data Scientist and ML Engineer | Document AI, Customer Analytics, LLM and RAG Evaluation | Python, SQL, PyTorch
- Handle
- @kartikey-r3jj
- Role
- Graduate Teaching Assistant
- Company
- George Mason University
- Location
- Fairfax, United States
- Open to
- Data Scientist, Data Analyst, Machine learning Engineer, Business Analyst
About
I’m a data scientist and machine learning engineer with 3 years of production experience, completing my MS in Data Analytics at George Mason University in December 2026 with a 4.0 GPA.
At Unifynd Technologies, I built and deployed ML systems end to end. My invoice recognition system, a graph neural network I trained from scratch in PyTorch, raised extraction accuracy from 48% to 78% on 10,000+ invoices a month and reduced manual review from 10 people to 3. I also built churn and retention forecasting, RFM customer segmentation, and recommendation models with XGBoost on 100,000+ customer records, and automated reporting dashboards that cut weekly reporting from 4 hours to 1.
Recently I’ve focused on LLMs: comparing GPT-4o against fine-tuned DistilBERT for named entity recognition with evaluation rubrics, building RAG pipelines with embeddings and vector retrieval, and creating LangChain agents.
Skills: Python, SQL, PyTorch, TensorFlow, scikit-learn, XGBoost, graph neural networks, NLP, LLMs, RAG, LangChain, A/B testing, causal inference, dbt, Airflow, Databricks, PySpark, AWS, Tableau, Power BI.
Open to Data Scientist, ML Engineer, and Applied AI roles in entire USA, or remote. Available January 2027. On F-1 OPT; eligible to STEM OPT, will need H-1B sponsorship.
Experience
- Graduate Teaching Assistant at George Mason University
Jan 2026 to now - Fairfax, United States
Hold weekly office hours for undergraduates, walking them through course concepts and assignments. Help the professor run and grade quizzes and assignments, and update course modules each semester. - Machine Learning Engineer 2 at Unifynd Technologies
Jan 2023 to Nov 2024 - Mumbai, India
Built an intelligent character recognition system for 10, 000+ customer invoices a month using a graph convolutional network trained from scratch, improving accuracy from 48% to 78%. Built Python and SQL data pipelines, cohort analysis, and XGBoost models for personalized recommendations and retailer campaigns. - Machine Learning Engineer 1 at Unifynd Technologies
Jan 2022 to Dec 2022 - Mumbai, India
Set KPI definitions for customer segmentation and built an RFM pipeline with K-Means and XGBoost to flag high-potential customers. Replaced a manual OCR performance report with Looker Studio dashboards and led development of an AI-powered chatbot POC leveraging SpaCy Natural Language Processing pipelines. - Data & Process Analysis Intern at Bosch
Jun 2019 to Apr 2020 - Bangalore, India
Analyzed spindle failure data and found early warning signs that reduced failures by 30%. Traced injector body damage to its root cause using Shainin methods, cutting assembly line defects by 70%, and built a Power BI attendance dashboard for 600+ machine operators.
Education
- George Mason University
M.S., Applied Information Technology (Data Analytics)
2025 to 2026
STEM OPT eligible. Expected Dec 2026. Relevant Coursework: Machine Learning, Statistical Analysis, Linear Algebra, Data Mining, Social Media Analytics - Manipal University
M.Tech, Industrial Automation and Robotics
2018 to 2020 - University of Mumbai
B.E., Electronics
2015 to 2018
Skills
- Python
- SQL
- MongoDB
- Pandas
- NumPy
- scikit-learn
- XGBoost
- PyTorch
- TensorFlow
- Hugging Face Transformers
- LangChain
- Power BI
Links
This profile as JSON