Hi, my name is
Nitigya Kargeti
I build intelligent systems.
I'm a Data Scientist and ML Engineer specializing in building (and occasionally designing) exceptional digital experiences. Currently, I'm focused on building accessible, human-centered AI solutions at UW-Madison.
With expertise in machine learning, natural language processing, and computer vision, I develop AI-powered applications that solve real-world problems and enhance user experiences.
Where I've Worked
Graduate Student Researcher
People and Robots Lab, NSF-Funded | Mar 2024 - Jan 2025
- Engineered an AI-assisted educational robot system (GPT-4o, specialized knowledge bases, triadic interaction model) resulting in a publication at ACM CHI 2025 ('SET-PAIRED') and demonstrating a 15% improvement in learning retention among 20 child-parent pairs (ages 3-4).
- Performed advanced statistical analysis using Wilcoxon Signed-Rank tests, revealing significant underestimation of children's capabilities in advanced math (p < .01) and phonological awareness (p < .05), providing accurate developmental insights to the parents.
Research & Development Intern
Centre for Development of Advanced Computing (CDAC) | Sep 2022 - Mar 2023
- Achieved a 30% reduction in system latency for the P-300 Keyspeller BCI system by re-engineering the signal processing pipeline with optimized implementations of ICA and DB6 Wavelet Template matching.
- Enabled reliable real-time BCI performance by implementing optimized epoching and buffering techniques for data stream processing, contributing to a 10% increase in keyspeller classification accuracy.
Software Development Intern
Spenza Inc | Remote, Bengaluru, India | Jan 2023 - Aug 2023
- Constructed a parallel PDF parser on AWS Lambda (AWS, Docker), processing 400K pages in 10s (80% faster) and reducing server costs by 30%
- Integrated Automation of Connected Accounts (Stripe API, Nest.js) to support both B2B and B2C models, boosting revenue by 18%
- Optimized 30% of payment logic (TypeScript) with multi-currency support and real-time conversion; enhanced error monitoring (Sentry) to cut resolution time by 40%
Research & Development Intern
CDAC | New Delhi, India | Sep 2022 - Mar 2023
- Implemented a real-time P300-Keyspeller (Python, NumPy, Pandas) with ICA (PyTorch) and DB6 wavelet matching, slashing latency by 30% and increasing alphabet classification accuracy by 10%
- Leveraged ensemble methods and epoching techniques (scikit-learn) for real-time processing, reducing latency by 35%
Summer Vocational Intern
DRDO | Dehradun, India | Jun 2022 - Jul 2022
- Analyzed hyperspectral satellite imagery (88% accuracy) using Linear Mixture Modeling and End-member analysis for terrain classification
- Streamlined Spectral Angular Mapper, cutting processing time by 10% and accelerating image workflows
Some Things I've Built
Chicago Crime Data Forecasting & Hotspot Analysis
Feb 2025
Engineered a hybrid ARIMA-LSTM predictive model that improved crime pattern forecasting accuracy to 85%, enabling law-enforcement to optimize patrol resource allocation and reduce response times. Deployed spatial clustering using DBSCAN to identify crime hotspots across 12 crime categories totaling 10M reports.
Quantitative Finance Risk Modeling
Jan 2025
Devised a multi-factor market volatility prediction framework that achieved 92% correlation with actual market movements. Developed a fully automated trading strategy with an integrated pipeline that ingests real-time data from Alpha Vantage API, reducing drawdowns by 40% in backtesting.
Hallucination Detection in Vision-Language Models
Dec 2024
Proposed an approach to identify hallucination sources in Large Vision-Language Models using Contrast-Consistent Search (CCS) across model components, detecting hallucinations with 86.7% accuracy compared to 53.8% baseline. Demonstrated hallucination prevention relies primarily on language modules, maintaining 75.2% accuracy under visual perturbation.
Sleep Apnea Detection
Mar 2023
Designed ApneaNet, a specialized CNN architecture, analyzing ECG signals to detect Obstructive Sleep Apnea (OSA), addressing a critical need for accessible screening methods beyond expensive sleep lab studies. Published findings in Biomedical Signal Processing and Control (IF: 5.7), receiving 19 citations to date.
What I Know
Programming Languages
ML/AI
Big Data/Cloud
Web Development
Get In Touch
I'm currently looking for new opportunities in AI/ML research and development. Whether you have a question or just want to say hi, I'll do my best to get back to you!