Muhammed Shabeer on Scaling Arré’s AI-Driven Short Audio Platform
Muhammed Shabeer, Technical Lead at Arré
Contact Sales
Technical Lead at Arré
Muhammed Shabeer is a seasoned software engineering professional and Technical Lead at Arré, where he spearheads the development of scalable, distributed systems powering the company’s audio-first social platform. With over five years of experience, he has a strong track record of building high-performance microservices and managing complex production environments across cloud platforms including AWS, GCP, and Azure, as well as on-premise infrastructure.
Before joining Arré full-time, Muhammed contributed as a backend engineer via Codemonk, where he focused on cloud-native services and NoSQL databases like Amazon DynamoDB and ArangoDB. He has also held engineering roles at Turbolab Technologies, Capcee Technologies, and co-founded LUD IoT Solutions, gaining hands-on experience in designing data pipelines, optimizing query performance at scale, and developing robust APIs and internal tools.
Muhammed is passionate about creating reliable systems that perform under pressure and is known for delivering practical solutions to complex engineering challenges.
Transcript
Arre Voice is basically audio social media platform where we make the content creation effortless and expressive. Unlike other platforms, we focus on audio where user can convey their emotions, their feelings authentically and also effortlessly.
Our mission is to empower users to share their thoughts, stories, creativity in the form of short audio. They don't need to worry about, storyboard or cameras, shoes or anything like that. So we are leveraging AI and we are one of the early adopters of generative AI in our tech stacks.
So, if we are focused on Indian market, there's an issue with different languages. We need, the modern Generative AI Models to understand the content better. Even If it's not in Hindi or English, we need to deal with different regional languages. And also for a social media platform, the moderation is crucial.
So we are leveraging AI to flag inappropriate content in our platform. DataStax powers our two critical systems. One is recommendation system. Vector embeddings are the backbone of our recommendation system which allow us to generate personalized recommendations in real time based on the user interest and the content as well.
And we are using it for conventional DB use case where it powers our creative analytics system. It's a rate heavy read light pattern. So, DataStax Astra DB is constantly delivering us single digit P99 latency.
Because of Astra DB serverless architecture, developers don't need to worry about, infrastructure management. They can focus on building features rather than, worry about all these things. And also Astra DB comes with the ability to serve both vector search and conventional DB use case, so we can use the same DB for both use cases.
We evaluated Milvus and Pinecone, but Astra DB stood out because of its smooth integration with the our scheme within our schema. And also because of all these serverless architecture, we don't need to worry about infrastructure or anything.
And that made us like 40 percentage cost savings within our, for our development. The AI is changing too fast. So, you need to focus on understanding your users deeply and build, or make their lives easier by leveraging AI.
Don't need to worry about the, operational overheads or infrastructure management. You need to focus on building things and making their life easier.