Milad
Bafarassat

AI for Wireless Communications

Developed CORTEX, a transformer-based beam selection system for terahertz (THz) UAV communications. The architecture combines convolutional front-ends with causal transformers using rotary position embeddings (RoPE) for drone state prediction. Key techniques include a two-stage approach for angle prediction and beam selection, circular-aware angle refinement, knowledge distillation, and focal loss for class imbalance.

Supervised by Prof. Özgür Gürbüz

Backdoor Attacks on Time-Series Models

Security evaluation and attack development for state-of-the-art time-series models. Implemented and extended two recent approaches: input-aware poisoning attacks that craft sample-specific triggers, and generative-model-based trigger synthesis (GAN / conditional generative models) to produce stealthy backdoors against RNNs, LSTMs, and transformer-based time-series classifiers. Conducted threat-model design, empirical evaluations on standard benchmarks, and initial defenses.

Supervised by Prof. Emre Özfatura

Mechanistic Analysis of Language Models

Researched mechanistic analysis and interpretability of transformer-based language models to enhance AI reliability and safety. Implemented a Meta-CoT-based approach to automate mechanistic interpretability in transformer architectures.

Supervised by Wenyu Du (HKU) and Dr. Jie Fu (Shanghai AI Lab)

FET Modeling Using Deep Learning

Developed robust FET models using data-driven techniques, exploring data augmentation strategies including GANs and autoencoders. Enhanced a limited dataset of 80,000 samples from 27 transistors for improved semiconductor device modeling.

Supervised by Prof. Korkut Kaan Tokgöz

LLM Agent Framework for Strategic Decision-Making

Conducted research on enhancing decision-making by LLMs in multi-agent settings by integrating external tools. Developed the STRIDE framework for strategic and interactive decision-making.

Supervised by Prof. Zhuoran Yang and Dr. Chuanhao Li

Political Affiliation Analysis Using CV and ML

Implemented self-supervised clustering algorithms (SwAV and Barlow Twins) to categorize over 7 million Twitter profile pictures, estimating the distribution of political affiliations and human/bot ratios.

Supervised by Prof. Onur Varol

AI Engineering Intern

• Designed and implemented a middleware pipeline connecting Unity game-engine events to LLM APIs.
• Developed a plugin-based architecture for extensible character and quest design.
• Produced documentation and performance benchmarks for real-time multiplayer scenarios.

AI Product Manager

• Led a cross-functional team of 7 software engineers and ML researchers plus 2 operations/marketing members.
• Co-founded and introduced 3 AI-driven products.
• Developed Recrupen, an ATS processing 5,000+ resumes annually for 200+ hires across 6 subsidiaries.
• Launched QShoot, an AI math problem-solving app reaching 10K users in 2 months.
• Launched Gorgeous, an AI makeup recommendation app reaching 1K users in 2 days.

Human Resources Manager

• Managed a team of 40 across HR, recruitment, L&D, and administration.
• Built and scaled a recruitment team from scratch to 7 full-time recruiters.
• Drove company growth from 100 to 300 employees within two years.
• Raised employee-satisfaction NPS from the low 80s to 92%.