AI Engineer - ASR: PraDigi, a Pratham Initiative

2 hours, 48 minutes ago
Full-time
Mid Level
Software Development

Description

  • Design representative multilingual speech-data collection and annotation processes.
  • Define transcript, reading-error, disfluency, and quality-control guidelines.
  • Identify data gaps and annotation inconsistencies through systematic model-error analysis.
  • Train, fine-tune, and evaluate ASR models for children’s read speech across languages and accents.
  • Build pipelines for audio preprocessing, segmentation, normalization, alignment, and model training.
  • Adapt architectures such as Whisper, wav2vec 2.0, Conformer, CTC, and RNN-T for reading assessment.
  • Optimize models for server, low-bandwidth, store-and-forward, offline, and mobile deployment.
  • Develop methods to detect reading substitutions, omissions, insertions, repetitions, hesitations, and self-corrections.
  • Translate model outputs into reading accuracy, WCPM, fluency, and reading-level measures with assessment and field teams.
  • Define ASR benchmarks, maintain reproducible evaluation pipelines, and document experiments.

Requirements

  • Degree in Mathematics, Computer Science, Data Science, or a related STEM field.
  • 3–5 years of full-time experience, including at least 2 years in an AI/ML role.
  • Strong experience with automatic speech recognition, speech processing, or spoken-language technologies.
  • Strong Python skills and experience with PyTorch or a similar deep-learning framework.
  • Experience with ASR architectures and toolkits such as Hugging Face Transformers, NeMo, Kaldi, or equivalent.
  • Knowledge of acoustic modeling, CTC/transducer/encoder-decoder architectures, decoding, beam search, language models, forced alignment, phonemes, WER/CER, and ASR error analysis.
  • Experience preparing and managing large-scale audio and transcription datasets.
  • Ability to diagnose model errors beyond aggregate transcription accuracy.
  • Availability for full-time employment with immediate joining.
  • Willingness to travel periodically to field locations in India.

Benefits

  • Flexible, remote-first work from anywhere in India.
  • Competitive, experience-based remuneration benchmarked against industry standards.
  • Opportunity to work on education technology supporting millions of children across 31 countries.
  • Exposure to advanced ASR and AI challenges in multilingual, low-bandwidth, and offline environments.
  • Autonomy to experiment with current state-of-the-art models within a mission-driven team.
  • Inclusive, collaborative workplace spanning Asia, Africa, and Latin America.

Interested in this position?

Apply directly on the company website

Apply Now

Similar Roles

Machine Learning Engineer II (ROA + LTV ML)

Affirm 1K-5K Diversified Financial Services

Affirm is hiring a Machine Learning Engineer in remote Canada to develop and scale production machine learning models that support credit risk and approval decisioning.

Apache Spark Deep Learning Feature Engineering Kubeflow Machine Learning NumPy Pandas Python Scikit-learn SQL XGBoost
1 hour, 18 minutes ago

Lead AI/ML Engineer

Arkestro 51-250 Internet Software & Services

Lead AI/ML Engineer at Arkestro, a company applying AI, game theory, and behavioral science to enterprise procurement, owning the design, productionization, and operation of AI-powered product functionality.

AWS CI/CD dbt Docker Git Kubernetes Machine Learning MLOps Python Snowflake
1 day, 1 hour ago

Machine Learning Engineer (Semi-Senior / Senior)

DaCodes 51-250 Internet Software & Services

DaCodes busca un/a Machine Learning Engineer Semi-Senior o Senior para integrarse remotamente al equipo de IA/ML de una startup argentina de ad-tech y construir modelos y soluciones inteligentes para optimizar la monetización de aplicaciones móviles.

Apache Airflow Apache Spark AWS Databricks Generative AI Machine Learning Pandas Python Reinforcement Learning Scikit-learn SQL Statistics
1 day, 2 hours ago

Senior ML Engineer

SSC Egypt 11-50 IT Services

Own production predictive customer scoring models for churn, propensity, lifetime value, spend intent, and response use cases, managing their full lifecycle within a self-managed data platform.

Machine Learning MLOps
2 days, 2 hours ago

You're on a roll! Sign up now to keep applying.

Sign Up

Already have an account? Log in

Used by 14,729+ remote workers