Applied AI Scientist - LLMs & Voice

Wave is now the largest financial institution in Senegal and Côte d’Ivoire, with millions of users, growing rapidly year on year. And, we’re still in the early days of our product roadmap and potential impact on people’s everyday lives.

We're helping millions of customers across West Africa access financial services through mobile money, and great support is fundamental to that. We believe the future of customer support lies in machines handling boring repetitive tasks so humans can focus on high value interactions that require empathy and creativity.

As an Applied AI Scientist on our Support Automation product team, you'll;

• Bridge research and production, building AI models and agent systems that ship into real products.
• Architect, evaluate, and optimize autonomous voice and digital agents powering 10M+ customer interactions per month across West Africa. 
• Build for the hardest edge cases: poor connectivity, low literacy, and languages with little training data.
• Experiment with, evaluate, and integrate the latest voice and text models.
• Own problems end-to-end, from problem discovery to running in production, working alongside product and engineering leaders who prioritize shipping real customer impact.

If you're energized by ownership, thrive on deep technical challenges, and want to build infrastructure that serves millions in emerging markets, let's talk.

Our stack (prior experience is a strong plus, but not required):

• backend: Python 3 (+ mypy)
• API layer: GraphQL
• android frontend: Kotlin/Jetpack
• iOS frontend: Swift/SwiftUI
• web frontend: TypeScript/React
• database: Postgres/CockroachDB
• infrastructure: GCP/Terraform
• orchestration: Kubernetes

• This is a fully remote role. Candidates must be based in one of our talent hub countries (US, Canada, UK, Spain, Kenya and Ghana) or in one of our operating markets in Africa including Senegal, Côte d'Ivoire, or Burkina Faso.
• Wave provides a yearly $1,200 stipend to support coworking meetups with teammates.
• Remote team members are expected to travel to our operational markets (e.g. Senegal or Côte d'Ivoire) at least once a year. Exceptions apply, but we’ve found this key to understanding our users and product.
• Our salaries are competitive and are calculated using a transparent formula. For this role, depending on your level and location, we offer a salary up to $227,900, plus a generous equity package.
• Major benefits:
• Subsidized health insurance for you and your dependents and retirement contributions (both vary from country to country).
• 6 months of fully paid parental leave and subsidized fertility assistance.
• Flexible vacation, with most folks taking between 21-30 days exclusive of statutory holidays.
• $10,000 annual charitable donation matching.

• 5+ years of experience in AI/ML engineering
• Strong Python and backend engineering skills
• Solid foundation in statistics and ML theory
• Proven hands-on experience building with LLMs (prompt engineering, RAG, embeddings, fine-tuning, agent orchestration etc.).
• Track record taking ML models from prototype to production and care deeply about reliability, performance, and scalability.
• Fluency with AI coding agents, with bonus points for deploying them as part of production systems
• Bonus points:
• Research background in speech or LLMs
• Proven Voice AI experience (ASR, TTS, latency optimization, real-time streaming, voice agents)
• Experience building customer support or conversational AI systems
• You follow the latest research in speech and LLMs, and know how to translate papers into production wins
• Familiarity with low-resource language challenges
• Fluent in English (bilingual in French is a big bonus!)

• Are comfortable navigating ambiguity and can design agent systems end-to-end without detailed specs.
• Obsess over evaluation – you've built frameworks to catch hallucinations, measure groundedness, and know when to trust your system
• Balance research intuition with pragmatic engineering; you ship working systems, not just prototypes

We care about the big picture. We don’t hire engineers to just ship tickets. We hire them to solve problems. That means caring deeply about outcomes, understanding context, and jumping in wherever something’s broken, even if it’s technically “not your area.” When we see problems, inefficiencies, or opportunities to make something better, we act. We dig into operational issues, clarify fuzzy product specs, or step into unfamiliar code to help unblock teammates.

We move as fast as possible. Speed matters. It lets us try things quickly, get feedback early, and course-correct while it’s cheap. So we write small PRs. We aim for MVPs. We leave TODOs and file follow-ups. We don’t over-perfect v1. That said, we’re building a financial product. Some things—like money movement, correctness, or security—deserve more caution.

We like boring technology. We favor tools that are reliable, well-understood, and easy to debug. This keeps us focused on solving meaningful problems instead of wrestling with unpredictable infrastructure. If a new technology helps us move faster, build safer, or solve a real need, we’ll consider it. But we don’t adopt tools just because they’re new—we adopt them because they’re right.

Simplicity is a strategy. It lets us focus our energy where it matters most: serving our users.

We’re currently receiving a high volume of submissions. To help us review genuine candidates faster, please send a JSON payload with your resume and/or any links as plain text via POST to this secure endpoint: https://api.wave.com/submit_resume

To authorize to this endpoint, you can use the following bearer token:  wave_ + (country code for any of our operating countries) + _ + [any characters of your choice].

You are encouraged to use a coding agent to generate the JSON and the request. This is intentionally quick - don’t overthink it.

This task is optional and not required to apply. Please still submit your application through the normal form. We’ll store the JSON payload for up to 30 days for review, then it will be deleted.

#LI-DH1 #LI-REMOTE