Erfan Afshar

Data Analyst

• Evaluated and improved AI-generated content with a focus on web search relevance, generative AI quality, and data validation.
• Analyzed and rated search results for relevance, language quality, and alignment with user intent to enhance search engine performance.
• Reviewed generative AI outputs (voice and text) for quality, providing actionable improvement suggestions.
• Cross-checked AI-generated content for factual accuracy using trusted external sources across multiple domains.
• Performed specialized evaluation tasks including visual content verification, intent classification, and color-based image analysis.

AI Automation Specialist

• Developed AI-powered voice automation systems for restaurants using N8N, Twilio, and OpenAI to enhance customer experience and reduce manual workload.
• Designed and deployed a Twilio + OpenAI voice assistant to handle real-time restaurant orders via natural phone conversations.
• Automated workflows for call handling, order logging, and customer data collection with third-party integrations.
• Implemented fallback flows for ambiguous AI responses, including human escalation options.
• Built personalized experiences for repeat customers with dynamic workflows suggesting past orders and responding contextually.

Research Assistant

• Completed thesis-based program 10 months early by optimizing research workflows and independently managing the project lifecycle.
• Designed and developed two novel LSTM-based models (MO and MIMO) for anomaly detection and cyberattack isolation on drone systems.
• Simulated control and cyberattacks (DoS, FDI, Replay) using MATLAB and Simulink; generated two large datasets (50K entries each) from quadcopter models.
• Achieved a 12% F1-score improvement over baseline models in identifying drone anomalies.
• Presented weekly updates to a 15-member research team and wrote a 130+ page technical report for industry collaborators.

Machine Learning Intern

• Built a preprocessing pipeline to clean, normalize, and transform user data from web activity logs (100K records) for ML readiness.
• Trained and fine-tuned ML models using Scikit-learn and TensorFlow to predict user behavior patterns with up to 90 percent accuracy.
• Evaluated and compared model performance using accuracy and runtime to select the most effective solution.

Software Developer Intern

• Developed Java-based REST APIs and microservices to automate workflows for robotic arm control, improving system responsiveness and maintainability.
• Integrated low-level C/C++ modules with Java services to enable real-time data exchange with industrial sensors and actuators.
• Contributed to a cross-functional engineering team by delivering scalable Java backend solutions that boosted automation reliability by 30 percent.

M.A.Sc. in Computer Engineering

Thesis: Anomaly detection, identification, and isolation in drones using enhanced LSTM-based models.
Focused on machine learning, neural networks, and control systems for autonomous drone security.
GPA: 3.9/4.3

B.Eng. in Computer Engineering

Covered core areas in AI, data mining, computational intelligence, cloud computing, software engineering, and embedded systems.
GPA: 3.6/4