Design and deliver production-grade AI/ML architectures for education and assessment platforms. Translate stakeholder requirements into secure, cost-aware blueprints covering RAG, multi-agent systems, LLM judging, OCR, human review, cloud infrastructure, LLMOps/MLOps, observability, governance, and evaluation. Lead workshops, architecture decisions, production readiness, FinOps, sovereign hosting, and technical mentorship across international government and education programs.

Develop and deploy multilingual automatic speech recognition systems for children’s reading assessment. Responsibilities span speech-data collection and annotation, audio preprocessing, model training and fine-tuning, decoding, forced alignment, error analysis, benchmarking, and deployment in low-bandwidth or offline environments. The role requires adapting modern ASR architectures, detecting reading errors, evaluating WER/CER and alignment quality, and collaborating with assessment and field teams to translate model outputs into reading fluency and accuracy measures.