Founded SwiftSignl and built JoBound around a hiring-market gap: most job tools surface listings after the highest-leverage application window has passed.
Built a scraping system across 20+ ATS integrations that normalized most jobs in under 2.5 minutes despite inconsistent schemas and rate limits.
Shipped a tool using CrewAI to help users improve their interviewing skills by using AI to debate the quality of their responses from their past interviews and provide feedback.
Built a low-cost pipeline to discover ATS customer lists programmatically, turning source expansion into a repeatable acquisition system.
Shipped the full platform, including a website, two apps, an API, and a local AI server, on near-zero spend; iterated directly from user observation and feedback.
Grew paying subscribers through word-of-mouth. Users reported more screening calls, and one early user landed a role within a month.
Working across multiple teams, introduced Sayari's first AI features - a RLHF system to improve the data pipeline quality and to summarize complex data for users.
Shipped two 0-to-1 products under investor pressure: Onbrand's PLM two months early, and Trayd's v1 ahead of YC Demo Day, contributing to a $15M raise.
With a team of 8, created v2 of the Qloo application to better support non-technical users who need access to their taste AI.
Turned around Qloo's delivery by stabilizing regressions, introducing TypeScript, and re-focusing planning, increasing throughput by 45%.
At Sayari, split a 2-repo monolith into 6 scoped sub-repos with shared libraries and API contracts, improving velocity and cross-team collaboration.