Mind Region

Logic Core

A systems-first lens on building AI that can reason, adapt, and execute with reliability.

This region captures my engineering approach to hybrid AI, reinforcement learning, and robotics-aware architecture designed for performance beyond controlled environments.

Structured and precise

Brain Map

Focus Areas

- AI-centric project design
- Problem decomposition
- Robotics-first thinking
- Scalable architecture

Highlights

- Built data-driven interfaces that turn messy information into guided decision flows.
- Prefer rapid prototyping followed by structured hardening for production readiness.
- Think in modules, feedback loops, and system constraints before writing code.

Technical Projects

AI Agents + Robotics

Hybrid Agentic AI for Autonomous Robots

Designed a hybrid AI architecture integrating reinforcement learning, neural perception, and symbolic reasoning to enable autonomous robotic navigation. Built agents using PyTorch and Unity ML-Agents, modeling reward functions for safe and adaptive decision-making. Implemented real-time perception and evaluated policy stability and convergence across dynamic environments.

PyTorch • Reinforcement Learning • Unity ML-Agents • Computer Vision • Search Algorithms

Applied AI Systems

AI for Airport Security: False Alarm Reduction

Developed transformer-based computer vision models to analyze X-ray and sensor data, reducing false alarms by 30%. Built real-time inference pipelines and optimized latency by 25% to support deployment under strict throughput constraints. Evaluated model performance using precision-recall analysis and ROC curves, achieving high detection reliability.

PyTorch • Transformers • Computer Vision • Edge AI

Human-Centered AI

Stutterio: AI Speech Fluency System

Built an AI-powered speech analysis system using deep learning and acoustic feature engineering to detect dysfluencies in real time. Designed pipelines extracting MFCCs, pitch variance, and temporal features, achieving 85% classification accuracy. Optimized inference to sub-200ms latency and incorporated privacy-aware on-device processing.

Python • Deep Learning • Audio Processing • NLP

Synchronizing neural interface...