Hyeonjoon Park

AI Engineering Intern

• •Led development of LUMINA, an AI-powered system enabling realistic voice-based conversations with digital replicas of influencers using advanced LLMs and voice cloning
• •Built a fully working PoC with average response time under 5 seconds, achieving 4.6/5.0 user rating for voice & style realism
• •Implemented dual-stage LLM architecture (Thinking LLM + Speaking LLM) for content reasoning and personality expression, improving modularity and style transfer
• •Developed comprehensive voice cloning pipeline using ElevenLabs Multilingual V2, achieving significant gains in voice similarity metrics (cosine sim, PCA, MFCC)
• •Established scalable credit-based business model with 77% gross margin and validated market potential of $21.9M annual revenue with 200K active users

Backend Development Intern

• •Designed and deployed a hybrid AI trading platform that combined LSTM-based time-series prediction and LLM-driven sentiment analysis to generate real-time, multi-asset trading signals on AWS
• •Engineered a custom vectorized backtesting engine (Python/C++) capable of simulating thousands of trades per second, including OHLCV replay, risk metrics (drawdown, Sharpe ratio), and strategy benchmarking
• •Trained LSTM models on historical market data with walk-forward validation, achieving 64% directional accuracy on 1-day asset returns

Division Commander Driver & Platoon Leader

• •Hand-selected to serve as the personal driver to a two-star general in a high-security division; executed confidential and time-critical missions requiring cross-unit coordination
• •Led and trained a 23-member transport unit in vehicle operations, mission planning, and safety protocols
• •Designed and implemented a C++-based driver assignment and mission scheduling system, automating over 1,000 logistics tasks; reduced manual overhead by 80%
• •Collaborated on experimental route optimization under variable terrain/military threat conditions using heuristic planning algorithms