Build on what
your team knows.
The people closest to a task know what good looks like. We’re building Orchestra to turn that judgment, and the work behind it, into intelligence they can keep.
The work is
the advantage.
Frontier models make new things possible. But a product’s recurring tasks, domain knowledge, and corrections are specific to the team building it.
Those lessons should not disappear after every request. They can become a clearer evaluation, a better prompt, a more capable route, or a specialist model trained for the job.
We start with evidence from the task. Keep the frontier where its capability matters. And make the improvements useful beyond the next model release.
How Orchestra worksThe founders
Product judgment.
Infrastructure depth.
Luis and Aamir worked together at Instacart. Orchestra brings their product and engineering experience to the work of making AI useful in production.
Co-founder & CEO
Luis Manrique
Previously employee number two at Gumloop, working across go-to-market, forward-deployed engineering, and data platforms.
Before that, Principal PM at Instacart on Ads and Carrot AI. His work spans applied machine learning, product development, and bringing new technology into enterprise workflows.
Co-founder & CTO
Aamir Poonawalla
Previously spent ten years at Instacart, most recently leading the Ads Serving and Infrastructure team as Senior Software Development Manager.
Earlier work includes Instacart’s Curbside Pickup initiative, Pinterest through the URX acquisition, and Microsoft. He holds a master’s in Computer Science from Georgia Tech.
Join Orchestra
Build what
comes next.
We’re looking for people who want to make AI work better in the real world. Tell us what you’ve built and what you’d like to work on.
Staff Infrastructure Engineer
Build the systems behind reliable model serving, routing, and training.
AI Research Scientist
Develop methods for evaluation, post-training, and learning from expert feedback.
Applied AI Engineer
Turn real workloads into better evaluations, prompts, and specialist models.
Forward Deployed Engineer
Work with customers to bring model improvements into production.
Product Engineer
Build the tools teams use to understand, improve, and deploy their models.
Sales
Help teams find where Orchestra can improve quality and reduce the cost of their AI workloads.