About the Role: Qubist Solutions is building a next-generation AI platform for the aviation sector. We are developing a highly specialized, mission-critical solution to automatically assess the radiotelephony (RT) communication between Air Traffic Control (ATC) and pilots.
We are looking for a Senior AI/ML Engineer to lead the architecture and development of an intelligent speech-and-text system that evaluates live and recorded VHF radio communications against strict international aviation standards (ICAO). If you love solving complex problems with noisy audio data, strict rule-based NLP, and real-time inference, this is for you
What You Will Build & Do:
Custom Speech-to-Text (STT): Develop and fine-tune acoustic models to accurately transcribe noisy, rapid ATC-pilot radio communications (both live streams and recorded files).
ICAO Compliance Engine:
Build an NLP/rules-based pipeline to evaluate transcribed text against prescribed ICAO Doc 9432 standard operating procedures and phraseology.
Deviation Detection: Engineer algorithms to identify omissions, incorrect terminology, mispronounced numbers, and non-standard conversational deviations.
Automated Scoring Logic: Design an objective, bias-free scoring mechanism that grades candidate communication for licensing and training purposes.
Real-Time Inference: Build low-latency audio processing pipelines capable of analyzing live communications securely on edge or on-premise government cloud environments.
Required Skills & Tech Stack:
Experience: 5+ years in Machine Learning, with a strong focus on Audio/Speech Processing and Deep Learning (NLP).
Audio/Speech Processing: Hands-on expertise with custom STT/ASR models (e.g., OpenAI Whisper, Kaldi, Wav2Vec, DeepSpeech) and handling noisy audio data.
NLP & LLMs: Robust experience with open-weight models (Llama 3, Mistra