Machine Learning Intern — LLM Post-Training & Multimodal Agents
Apple Inc. • Jan–Aug 2026
Building large-scale multimodal generation pipelines: LLM-orchestrated synthesis of audio, 3D scenes, and spatial environments via multi-agent coordination. Developing controllable generation systems with LLM planning and diffusion/codec rendering for production-scale quality. Post-training LLMs for generation tasks, aligning multimodal agents to coordinate specialized synthesis modules via tool-calling APIs.
Research Scientist Intern — Audio GenAI
Smule AI • Jun–Sep 2025
Developed spatial audio generation capabilities with prompt-based control for real-time applications using modern generative modeling approaches. Built training infrastructure for multimodal audio synthesis models from scratch using distributed computing frameworks.
Research Intern — Audio/Video/LLM
Microsoft Research • May–Aug 2024
Developed coordination system for multiple MLLMs and LLMs to extract unified features through cross-modal attention between audio and video. Built multi-LLM pipeline for large-scale audiovisual learning using distributed computing frameworks. Research on unified audiovisual encoder accepted at EUSIPCO 2025.
Audio–Language Research Intern
Bosch AI Research (BCAI) • Apr–Jul 2023
Co-authored counterfactual audio–language method (ICASSP 2024). Developed novel approaches for learning audio concepts from counterfactual natural language prompts.
Patents & Intellectual Property
Method and System to Train Audio Retrieval and Zero Shot Classification Systems with Counterfactual Prompts
US Patent US20250124292A1 • Published January 2025
Assignee: Bosch AI Research USA
Ali Vosoughi 