# Kiran Prasad J. P.

> Data Scientist & Application Developer | Expertise in Deep Learning (PyTorch/TF), Reinforcement Learning, Computer Vision, Healthcare AI and Production Flutter (GCP, Docker) | Developing Innovative Data Solutions

Location: Pittsburgh, Pennsylvania, United States
Profile: https://flows.cv/kiranprasadjp

Lead Data Scientist and Flutter Developer at Carnegie Mellon University. My role exists at the intersection of "Deep Math" and "User Experience"— researching complex AI models; Building the robust, full-stack ecosystems required to deploy them. Previously, as a Graduate IT Associate at CMU, laid the groundwork for my operational expertise, troubleshooting enterprise hardware/software ecosystems and streamlining user management enhancing efficiency.

Graduated from Carnegie Mellon University, holding a Master's degree in Artificial Intelligence Engineering - Chemical Engineering. This background allows me to bridge the gap between physical systems and digital intelligence, comfortable navigating both thermodynamic equations and advanced neural network architectures (CNNs, Transformers, LSTMs).

My technical philosophy is best illustrated by the University Seminar Attendance System I architected using Flutter and Firebase. 
- Fraud Detection: I engineered a multi-factor validation pipeline utilizing geofencing and outlier detection to identify and neutralize sophisticated fraud attempts, specifically GPS spoofing.
- Automation: The system streamlined certificate eligibility, eliminating manual oversight for administrators and securing the process for students.

I am an Oracle Cloud Infrastructure 2025 Certified Professional (Data Science & Generative AI), and my portfolio spans the bleeding edge of intelligent systems:
- Generative AI: Denoising Diffusion Probabilistic Models and Generative Design of Airfoils using VAEs/GANs.
- Reinforcement Learning: Solving Vehicle Routing problems with Graph Attention Networks.
- Computer Vision: Adversarial attacks on self-driving cars (SafeBench) and End-to-End Chest Cancer Classification.
- Engineering AI: PDE-GPT (Solving Physics with Transformers) and Fault Detection in Batch Reactors.

I thrive on problems that make others scratch their heads. Whether it's breaking into a secure system (ethically!), debugging a stubborn Adversarial Network, or pursuing perfection with a vengeance—yes, I’ll refactor code if it’s not elegant—I am a puzzle-solver at heart.
I’m a gamer (shoutout to Kratos and Geralt), an artist (sketching on Procreate), and a high-functioning problem solver with a passion for deduction. When I'm not coding, you'll find me literally cooking—ask me about my signature dish.

## Work Experience
### Data Scientist | Application Developer (FLUTTER) @ Carnegie Mellon University's College of Engineering
Jan 2025 – Present | Pittsburgh, Pennsylvania, United States
The goal was to create a seamless, automated digital solution that ensured high data integrity without compromising user experience.
I architected a robust Flutter application integrated with a Google Cloud Platform (GCP) & Firebase backend. The system replaces manual sign-ins with a secure, event-specific 6-digit code validation system.
- Eliminated proxy attendance and ensured user privacy by developing a Flutter application that executes Geofencing verification locally on the client-side, ensuring zero raw location data is transmitted to the server.

- Optimized user experience, achieving 35% faster check-in times by implementing A/B testing via Firebase Remote Config to dynamically toggle Geolocator settings based on real-time network conditions.

- Reduced false-negative validation errors by 12% by conducting rigorous Hypothesis Testing to calibrate the geofence radius, balancing security requirements with GPS hardware limitations.

- Neutralized GPS spoofing attacks by engineering Automated ETL Pipelines that feed into an Outlier Detection model, identifying spatial anomalies in "successful" check-ins that bypassed initial validation.

- Enabled scalable, zero-downtime event management by architecting a hybrid data model using GCP Firestore for real-time syncing and Cloud SQL (Firebase)for structured historical archiving.

- Streamlined institutional onboarding by integrating Federated Authentication with University SAML SSO, reducing login friction while maintaining strict access control standards.

### Data Scientist | IT Support Assistant @ Carnegie Mellon University's College of Engineering
Jan 2025 – Jan 2025 | Pittsburgh, PA
- Eliminated manual data entry bottlenecks and reduced processing lag by 40%, achieved by engineering an automated ETL pipeline to synchronize high-volume key check-in/check-out datasets.

- Restored data integrity and recovered 18% of incomplete records by executing advanced data wrangling, utilizing temporal correlation algorithms to impute missing User IDs.

- Unlocked granular event analysis and user segmentation capabilities by transforming unstructured raw logs into structured, schema-enforced formats using Pandas DataFrames.

- Secured 95% data cleanliness and enabled reliable downstream predictive analytics by establishing a rigorous Data Quality Framework to filter noise and validate raw inputs.

### Graduate-IT Associate @ Carnegie Mellon University's College of Engineering
Jan 2024 – Jan 2025 | Pittsburgh, Pennsylvania, United States
In my role as a Graduate-IT Associate at Carnegie Mellon University's College of Engineering, I excelled in troubleshooting hardware and software issues, managing user accounts, and utilizing various IT tools to enhance operational efficiency. 

• Demonstrated effective collaboration in team environments through leadership roles in academic projects and extracurricular activities.
• Leveraged strong analytical skills and passion for technology to deliver exceptional user support and contribute to streamlined IT operations.

### Graduate Researcher @ Carnegie Mellon University's College of Engineering
Jan 2023 – Jan 2024 | Pittsburgh, Pennsylvania, United States
Delivered a 10% increase in operational efficiency and minimized logistics costs for Linde Inc. by engineering a Deep Reinforcement Learning framework (GAT + RL) to optimize complex vehicle routing and load balancing tasks.

Outperformed Google OR-Tools and Deep Q-Networks (DQN) in computation speed while achieving 62% routing efficiency by refining the RL agent with Policy Gradient methods and Adaptive Sampling techniques.

Optimized spatial encoding and route prediction on high-dimensional data by implementing Multi-Head Attention mechanisms and Node Embeddings within a Graph Attention Network, utilizing Data Augmentation to improve model generalization.

Enhanced model robustness and prediction accuracy by deploying Ensemble Learning and Gradient Boosting strategies to handle dynamic constraints and varying route topologies.

### PLUS Tutor @ PLUS - Personalized Learning Squared
Jan 2024 – Jan 2025 | Pittsburgh, Pennsylvania, United States
- Mathematics Tutor at PLUS - Personalized Learning Squared, affiliated with Carnegie Mellon University's School of Computer Science. 
- Developed personalized math tutoring techniques for students from diverse backgrounds to enhance skills and confidence.
- Utilized interactive tools to create engaging learning experiences and inspire interest in math and STEM fields.
- Provided tailored feedback and guidance to help students achieve academic goals.

### Head of Design @ IIChE BMSCE Student Chapter
Jan 2021 – Jan 2023 | Bengaluru, Karnataka, India
In my role as the Head of Design at IIChE BMSCE Student Chapter in Bengaluru, Karnataka, India, I piloted design initiatives for college posters and magazines, garnering recognition from management. I orchestrated design strategies, cultivated a culture of creativity, and executed exceptional design solutions using tools such as Adobe Photoshop and InDesign. Additionally, I designed and published a college magazine, ChemZone-2023, showcasing my ability to deliver high-quality design projects.

### Machine Learning Intern @ Indian Institute of Chemical Engineers
Jan 2022 – Jan 2023 | Bengaluru
During my internship at the Indian Institute of Chemical Engineers, I led an end-to-end machine learning project tackling a real operational problem in renewable energy: solar panel defect detection. Large installations require regular inspection for surface issues like electrical damage, physical cracks, dust, and bird droppings — but manual visual inspection is slow, inconsistent, and difficult to scale without technical expertise.
I owned the full pipeline from day one. My initial custom 11M-parameter CNN revealed a textbook overfitting problem — 99% training accuracy against just 61% on unseen data — so I pivoted to a transfer learning approach, benchmarking MobileNetV2 and EfficientNet before selecting EfficientNet for its superior feature extraction capability. After systematic hyperparameter optimization to stabilize training, I deployed the final model as an interactive web application using Streamlit, deployed the entire pipeline to an AWS EC2 Ubuntu Linux server for continuous, real-time access via nohup with Git managing version control throughout.
The optimized model achieved 83.05% validation accuracy across 6 defect classes — a 22-point improvement over baseline. Non-technical field staff can now upload panel images and receive instant health diagnostics with top-3 confidence breakdowns, transforming what was once a labor-intensive manual process into an accessible, real-time diagnostic tool.

### Engineering Intern Student @ Donimalai Iron Ore Mine of NMDC Limited
Jan 2022 – Jan 2022 | Ballari, Karnataka, India
Collaborated with senior engineers and process team members to analyze manufacturing processes and identify inefficiencies in ore processing. Collected and analyzed production data using Python and SQL and checked for improvements in production. Contributed to refine and upgrade ore quality and grade.

### Internship Trainee @ Karnataka Soaps and Detergents Limited (Mysore sandal soap factory)
Jan 2022 – Jan 2022 | Bengaluru, Karnataka, India
Co-operated with process engineers to improve the efficiency of the process. Gathered operational data to find ineffectiveness in process steps using SQL and visualized data using Power BI. Organized and participated in factory site visits. Drafted a final report indicating improvements in the production of soaps and cosmetics.

### Student Intern @ Karnataka Cooperative Milk Producers'​ Federation Limited
Jan 2022 – Jan 2022 | Kolar, Karnataka, India
Hands-on experience with fully automated and advanced technology in ultra-high-temperature treatment. Gained experience in different operations in milk processing, Quality control, packaging, and logistics. Implementation of clean production and zero liquid discharge was an exhilarating experience.

### Engineering Intern @ Indian Institute of Chemical Engineers
Jan 2022 – Jan 2022 | Bengaluru, Karnataka, India
- Performed an Support Vector Machines (SVM) based anomaly detection system to identify defects and irregularities in production line data with 87% accuracy. Leveraged sensor data for model training, improving product quality and cutting down waste by 12%.
- Conducted thorough data preprocessing and feature engineering to enhance SVM model performance by 8%. Collaborated with cross functional teams to integrate system into existing workflows, ensuring seamless adoption and operational efficiency.


## Education
### Master of Science - MS in Artificial Intelligence Engineering and Chemical Engineering
Carnegie Mellon University

### Bachelor of Engineering - BE in Chemical Engineering
B. M. S. College of Engineering


## Contact & Social
- LinkedIn: https://linkedin.com/in/kiranprasadjp
- Portfolio: https://captainklsh.github.io/Portfolio-KJP/

---
Source: https://flows.cv/kiranprasadjp
JSON Resume: https://flows.cv/kiranprasadjp/resume.json
Last updated: 2026-04-17