Engineering leader specializing in data platforms, analytics infrastructure, and AI-native applications.
I build the systems that help companies make better decisions with data—from mobile instrumentation processing billions of events to LLM-powered visualization tools.
- Leading platform engineering at Preset for Apache Superset—scalability, reliability, and AI integration
- Building production MCP (Model Context Protocol) integrations for natural language data visualization
- Architecting AI testing frameworks and observability infrastructure for enterprise data platforms
- Advising startups on analytics infrastructure and engineering org design
- Current: Staff Software Engineer at Preset (Platform & AI Integration)
- Previously: Engineering Leader at Tinder (10x engineer → 10x manager: built Hubble/EventsSDK analytics platform, scaled team 7→25 engineers)
- Before that: Senior Engineer at Uber (built Piper workflow platform, one of first mobile engineers on Rider Growth)
- Also: CTO & Cofounder at Alvear Lending (fintech platform)
- Roots: Architected XDMoD, an NSF-funded open-source HPC analytics platform still operating at national labs 14+ years later
languages = ["Python", "Kotlin", "Java", "Go", "SQL", "TypeScript"]
frameworks = ["FastAPI", "Flask", "Spring", "gRPC", "React", "Android"]
data = ["PostgreSQL", "Redis", "Kafka", "Hive", "Protobuf", "Celery"]
cloud = ["AWS", "GCP", "Docker", "Kubernetes", "Terraform"]
observability = ["Datadog", "OpenTelemetry"]
ai_ml = ["LangChain", "FastMCP", "Pinecone", "FAISS"]- Data Engineering: Event-driven pipelines, analytics platforms, instrumentation
- Engineering Leadership: Team scaling, org design, technical strategy
- Mobile + Backend: Android, iOS, gRPC/Protobuf, microservices
- AI Integration: MCP tools, LLM-driven applications
- Regulated Systems: Fintech compliance (KYC/AML, SEC/Reg CF)
- 📧 [email protected]
- 📍 San Francisco, CA
- 🌐 aminghadersohi.com (coming soon)
- 10x engineer → 10x manager at Tinder: Built Hubble/EventsSDK analytics platform (billions of events/day, 98% data quality), then scaled team 7→25 engineers, interviewed 200+ candidates, promoted 8 engineers
- Built Uber's Piper workflow platform from scratch—now powering 200,000+ pipelines across 1,000+ teams (grew 2,600x from initial 75 workflows)
- One of Uber's first mobile engineers—architected Thrift-based content platform serving millions of marketing impressions daily
- Achieved 95% first-attempt success rate for LLM-generated visualizations at Preset
- Architected XDMoD, an NSF-funded HPC analytics platform still running at national labs 14+ years later (93 stars, 4,666+ commits, v11.0.2 released Aug 2025)
Building platforms that turn data into insights since 2010