I build reliable APIs and intelligent systems that ship.
Java Spring, FastAPI, PyTorch and DevOps for production teams.
Full-stack engineer and AI researcher turning complex backend, MLOps and traffic-intelligence problems into fast, maintainable products.
My Technical Skills
Backend, AI/ML, DevOps, Frontend
4+ Years Production Expertise
Backend Engineering
Artificial Intelligence
DevOps & Infrastructure
Modern Frontend
My Technical Projects
Backend, AI/ML, DevOps in Production
Architecture, Code, Concrete Results
Secure Management API
Microservices + Kafka
High-Perf HTTP Server
Complete CI/CD Pipeline
End-to-End MLOps Pipeline
Computer Vision (AI)
AI Research & Master's Thesis
Unified Traffic Perception & Prediction
YOLOv8 + ST-GCN • Weather-Adaptive Fusion
Unified Traffic Perception & Prediction for Adverse Weather Conditions
Intelligent Transportation Systems (ITS) rely heavily on deep learning, but existing models suffer from brittleness in adverse weather (fog, night) and siloed operation. I developed a novel, real-time framework integrating YOLOv8 object detection with ST-GCN spatiotemporal forecasting. The core innovation is a bidirectional, weather-adaptive fusion engine that dynamically balances models based on environmental conditions, achieving a 24.3% performance improvement.
class WeatherAdaptiveFusion(nn.Module):
def __init__(self):
super().__init__()
self.alpha_det = nn.Parameter(torch.tensor(0.5))
self.alpha_pred = nn.Parameter(torch.tensor(0.5))
def forward(self, det_features, pred_features, weather_context):
# Dynamic weighting based on weather conditions
w_det, w_pred = self.get_adaptive_weights(weather_context)
# Bidirectional feature fusion
fused_state = w_det * det_features + w_pred * pred_features
return fused_statePerformance Under Adverse Weather
Fusion Architecture
ST-GCN Network
Scientific Publications
IEEE Conferences, Impact Factor Journals
ST-GCN, YOLO, Traffic Analysis
A Multi Stage Ensemble Framework for Fake News Detection: Integrating Traditional Machine Learning, Deep Learning, and Advanced Feature Engineering
Published on Mar 27, 2026, pages 2802-2817. This journal article presents a four-stage ensemble framework for fake news detection that combines traditional ML, deep learning, and advanced feature engineering for robust binary classification.
Enhanced Deep Learning Models for Real-Time Traffic Analysis
Comprehensive study on joint optimization of ST-GCN and YOLO for real-time traffic analysis under adverse weather conditions.
Evaluation of Deep Learning Architectures for Traffic Sign Classification
Quantified 18.5% generalization gap across 10 architectures (ResNet, ViT, EfficientNet) and 4 datasets. Benchmarked adversarial robustness (PGD/FGSM) and hardware efficiency (Edge vs Server).
Leak-Free ML for Real-Time Urban Crash Severity Prediction
Engineered a leak-free pipeline with 124 features from 166k NYC records. Achieved 0.960 Macro-F1 and <0.1ms latency using cyclical temporal encoding and spatial clustering.
About Medindev
Backend Engineer & AI Researcher
Languages
Code that bridges continents & disciplines
Results-driven backend engineer with expertise in Java/Spring Boot, Python FastAPI, Node.js/Express, C/C++, and cloud-native deployment. MSc Computer Science graduate (Nanjing) with 4+ years building scalable APIs. Open to remote jobs and new projects.
Born and raised in Gabon, I left my home country in 2018 to explore the world. Educated across cultures with coursework in South Africa and China, I bring a multilingual mindset to engineering: French precision, English pragmatism, and a dash of Mandarin patience.
My Journey
Core Interests
Curriculum Vitae
Experience, Education, Certifications
Direct PDF Download
A4 Format • 3 Pages • English/French
Experience
API Developer & Tech Specialist
APIDog • 2025-Present
Content & SEO Specialist
QUICK DOGTHIS • 2024-2025
Founder & Tech Lead
MedinChina Sourcing • 2020-Present
Education
Master's in Computer Science
Nanjing University of IST 2024-2026
B.S. Computer Science
Sanming University 2020-2024
Certifications
CV updated March 2026
Contact for OpportunitiesOpen to Remote Jobs & Projects
Backend • AI/ML • DevOps • Custom Projects
Available for new opportunities • Response within 24h