Machine Learning Research Scientist (ASR) Remote

Machine Learning Research Scientist (ASR) RemoteFormal position title: Machine Learning Research Scientist (Speech & Language Technologies)Job Overview: Intron is a health-tech startup building ASR for clinicians and patients with African languages and accents, helping reduce burnout from tedious documentation for already overworked doctors across Africa. Clinical Speech Recognition is ubiquitous in developed countries but virtually absent across Africa or underperforms with African accents. You can help fix that! Through the contributions of over a thousand African clinicians, we have created Africa's largest and most diverse clinical speech datasets (10,000+ hours, 300+ accents, and 23 languages). Come put your superpowers to work doing ASR, TTS, Clinical NER, Machine Translation, and Generative AI!Employment type: Compensation is based on experience, qualifications, cultural fit, and the number of hours available per week.ML Research Scientist (Full-time, Part-time): Remote, up to $5,000 full-time, up to $2,500 per mo part-timeML Research Graduate Intern (PhD/MSc, Full-time, Part-time): Remote (Part-time or Full-time) for 3-6 months. Up to $2,000 per mo full-time, up to $1,000 per mo part-timeWho You Are:Fluent with PythonMachine learning and deep learning research/experience are required.Experience should go beyond basic fine-tuning, inference, and API calls to explore making architectural or algorithmic modifications to models, loss functions, optimizers, etc.Understand the math and core concepts that underly deep learning, e.g. gradient descent, momentum, regularization, overfitting, bias-variance, etcResearch, projects, or publications in Automatic Speech Recognition and/or Natural Language Processing are required.Experience working with deep learning frameworks (Pytorch, Tensorflow, etc.) and related tools.Experience with LLMs, conversational agents, text generation, or summarization is a plus but not requiredExperience with distributed training with multiple GPUs in the cloud (AWS/GCP/Azure) is a plus but not requiredUnderstands the theory and practice of the Design of Experiments and statistical analysis of results.Comfort collecting, manipulating, combining, and analyzing complex, high-volume, unstructured data from varying sources.Familiar with the techniques and limitations of observational studies.Publication record in top ML/DL conferences/workshops is highly desired but not required.We are looking for people who are:Growth-oriented: Strong desire to push your ideas into production, overcoming obstacles, ready to learn, and grow your skills to match any challenge.Personable and Fun: You are great to work and talk withUser-focused: You are passionate about deploying models and applications that improve the lives of millions of users and enjoy research that has direct user impact.Communicative and collaborative: You are able to work effectively with others. You will be working closely with other researchers, engineers, designers, product managers, user experience researchers, and customer groups to build product features and high-quality products.Nimble: Balancing short-term execution with longer-term concerns, can pivot when necessaryOwnership: You have demonstrated feature development ownership, and can take initiative and responsibility for building, shipping, and maintaining core features, end to end.How to apply: Please fill out the application form and upload your resume by clicking the "Apply" button below.Your resume should be in reverse chronological order showing your most recent experience/projects first. Under each experience, kindly provide 3 to 4 bullet points describing interesting problems you solved, achievements, or important lessons learned on the job. The Education/Academic qualification section should follow. Next, the achievements/awards section could follow. Lastly, include any other information you think may be RELEVANT to the role.Tobi Olatunji is a physician turned Machine Learning Scientist with a passion for Global Health.Tobi's journey to starting Intron is nothing short of epic. He's worked with big names like AWS Health AI, Enlitic Inc., and Cambia Health Solutions building intelligent Natural Language processing tools for large health systems in Australia, Brazil, Canada, Japan, UK, and the US. He's on the Advisory Board at Harvard's OpenNotes Lab, pushing for responsible AI in healthcare. His research spans accented speech recognition, algorithmic bias, and more, with publications in top machine learning conferences like NeurIPS, EACL, Interspeech, and EMNLP.With a mix of medical and tech degrees, including an MBBS, an MSc in Medical Informatics, and MSc in Computer Science at Georgia Tech, a certificate in Healthcare Management from Yale School of Management, plus three US tech patents, Tobi is a true innovator. He's also a key member of the Commonwealth AI Consortium and the Research Director at Bio-RAMP Labs.Google Research, Gates Foundation, and Intron collaboration: linkCommonwealth Partnership Announcement: link and linkHarvard+BIDMC OpenNotes Advisory Board: linkHarvard Radcliffe Consortium on scaling Digital Health Innovations in Africa: linkCool research papers coming out of our lab:AfriSpeech [MITPress 2023]: our pan-African clinical speech dataset with benchmarksAfriNames [Interspeech 2023]: most ASR models butcher African namesAccentFolds [EACL 2024]: models can learn African cultural and geopolitical relationships from speech dataASR on Medical Entities [Interspeech 2024]: models can work great on non-medical speech but perform worse on medical termsAfroTTS [Interspeech 2024]: A 1000 African Voices; generating English speech in nearly 1000 African accentsAfriMed-QA [ACL 2025*]: the largest study on LLMs in African healthcareMachine Learning Research Scientist (ASR)J-18808-Ljbffr