S
Start.io operates a sell-side advertising platform that processes hundreds of millions of ads daily across mobile apps and websites, powered by real-time audience data and ML inference. The tech stack reveals a production ML infrastructure heavy on inference (Triton, TorchServe, BentoML) and streaming (Kafka, Pub/Sub), with active projects spanning low-latency inference, feature stores, and LLM/RAG systems—indicating a shift toward AI-driven campaign optimization and real-time decisioning. Hiring velocity is accelerating across data (3 roles) and sales (3 roles), suggesting both platform maturation and sales expansion.
Tech Stack 30 technologies
Core StackPython FastAPI Kafka Kubernetes AWS Redis Prometheus Grafana OpenTelemetry Docker PySpark MongoDB Jira Triton Inference Server TorchServe BentoML Pub/Sub GCP OCI Aerospike Bigtable SQL AWS SQS Vertica Testmo TestRail XGBoost LightGBM Kibana Excel
What Start.io Is Building
◆Challenges
- Low-latency inference services
- High availability scalability observability
- Optimizing latency throughput cost-efficiency
- Meeting revenue targets
- Campaign issue resolution
▲Active Projects
- Low-latency inference services
- Real-time decision-making engines
- Model versioning shadowing a/b testing
- Batch data pipelines
- Real-time nlp applications
- Llm and rag systems
- Real-time production algorithmic solutions
- Low-latency feature store
- Scalable data pipelines
- Real-time data models
Hiring Activity
9 roles · 4 in 30d
Department
Seniority
Company intelligence
