Salsabil Maulana Akbar

AI/ML Engineer · Speech AI · NLP · Information Retrieval

Kuala Lumpur, Malaysia

AI/ML Engineer with 6 years of experience building production systems across Speech, NLP and Information Retrieval, Computer Vision, and Tabular ML. Experienced across the ML lifecycle, from research and experimentation to data pipelines, model optimization, deployment, and observability.

GitHub LinkedIn Hugging Face Google Scholar Semantic Scholar Email

🚀 Core Experience

Production Systems

Built property valuation models, improved search relevance, and deployed multilingual speech models and scalable ML pipelines into production.

Experiment Design & Impact Measurement

Designed evaluation setups and measured the quantitative impact of model changes on relevance, conversion, and key business metrics at scale.

Generative AI & Voice

Currently researching, fine-tuning, evaluating, and deploying multilingual ASR and TTS models alongside training and benchmarking platforms for production voice systems.

Research

Published at EMNLP, ACL, and AACL on multilingual and low-resource NLP.

🛠️ Tech Stack & Where I Used It

Voice AI (current role): Python, PyTorch, Hugging Face, NeMo, FastAPI, vLLM, Triton, NVIDIA NIM/Riva, LiveKit, AWS, Docker, Kubernetes, DVC, GitHub Actions
Search & IR (Tokopedia, TikTok): Query understanding, graph embeddings, retrieval & ranking, vector databases, model distillation & quantization, TensorFlow, SQL, Airflow, GCP
Proptech (Pashouses): Tabular and geospatial modeling, computer vision, OCR, ETL pipelines, PostgreSQL, GCP
Mathematics & teaching: Applied mathematics, statistics, MATLAB, R

🔭 Currently Exploring

Trustworthy AI

Actively growing into Explainable AI and Causal Inference, working toward AI that is trustworthy and interpretable.

📄 Selected Publications

All papers & awards →

Education & experience

Activities

Teaching & talks