| Location: Chattanooga, TN Remote: Yes Willing to relocate: Yes Technologies: Python, PyTorch, Whisper/Ollama, Local LLMs, Research Synthesis Résumé/CV: https://taha-y-merghani.github.io/Taha_Merghani_Resume.pdf Email: taha.y.merghani@gmail.com Research Engineer who ships.
I bridge the gap between academic research and production systems. I hold an MS CS (Georgia Tech) with published research at NeurIPS/NAACL, but my focus is on building things that work in the real world and understanding why they break. My writing on human expertise and AI hit #1 on r/ArtificialIntelligence (>200k views). My essay on my own resilience was shared by Jeff Dean. What I build:
End-to-End Local Voice AI: A privacy-first voice assistant running locally using Whisper + Ollama + Mistral. Full-stack integration of audio and LLM modalities on consumer hardware.
Automation from the Inside: I currently work retail while rebuilding my career. I wrote a technical analysis of why store automation fails, examining sim-to-real drift, impedance mismatch, and latent provenance. Not from a lab. From the floor.
Research training under prominent AI researcher. I know how to go from paper to experiment design. What I'm looking for:
A role in ML Engineering, Applied Research, or LLM Evaluation where the job is to take cutting-edge systems and make them robust in the real world. I'm interested in the seams, where models meet reality and break. Whether that's building or stress-testing, I want to be close to the failure modes. |