• Nov 2024 - Present
    Founding Engineer

    Avoca is a YC-backed AI platform for home service businesses such as HVAC, plumbing, and electrical. I joined at seed stage as an early engineer after inbound call handling drove initial traction, then built voice and text booking agents and the outbound product from the ground up as the company scaled to Series B (~35x growth).

    Key contributions

    • Built Avoca's outbound platform from the ground up after inbound drove initial customer wins; outbound now drives ~30% of revenue as the company scaled from seed to Series B (~35x growth).
    • Built voice and text booking agents that use tools for availability, pricing, customer history, fees, and scheduling to automate job booking and CRM record updates for home service customers.
    • Shipped a self-serve workflow engine with call and text nodes and delay edges so customers could run follow-up, reminder, and reactivation campaigns without engineering support.
    • Implemented throughput controls, scheduling, and compliance guardrails to run high-volume outbound voice and text without breaching rate or regulatory limits.
    • Interviewed and hired engineers; mentored the team through design and code reviews.
    • Owned production reliability with PostgreSQL query optimization, monitoring, and stability work across core communication infrastructure.
    • Worked directly with customers to gather requirements and ship outbound capabilities tailored to their campaigns and operations.

    Technologies

    TypeScript, Node.js, PostgreSQL, Supabase, AWS, Twilio, OpenAI, Inngest, Vercel

  • July 2019 - Dec 2022, Jun 2023 - Nov 2024
    Principal Software Engineer

    Fyle is a fintech SaaS company for expense management, later acquired by Sage. Across two tenures, I advanced from Member of Technical Staff to Principal Software Engineer while the company grew roughly 25x in revenue from Series A to pre-acquisition. I led a platform initiative to rewrite the backend for 150+ APIs with dual read/write-optimized schemas and a migration path that kept legacy and new APIs consistent, laid foundation for pgevents, an open-source Postgres CDC library that streams logical replication events to RabbitMQ, then onboarded the broader engineering org to ship confidently on the new stack.

    Key contributions

    • Led a platform initiative rewriting 150+ APIs onto read/write-optimized Postgres with custom CDC, enabling a ~40-engineer org to migrate incrementally while legacy and new clients kept shipping without disruption.
    • Standardized new GET APIs on PostgREST-style query patterns, a custom parser, and query-layer security with bidirectional legacy sync so teams could ship consistently.
    • Built an internal assistant from tickets, knowledge bases, video, and Slack to shorten customer success onboarding and reduce time-to-answer for customer-facing teams.
    • Shipped API, event, and SQL function test coverage with local fixtures and GitHub Actions, raising release confidence as the company grew roughly 25x in revenue from Series A to pre-acquisition scale.
    • Laid foundation for pgevents (github.com/fylein/pgevents), streaming Postgres logical replication events to RabbitMQ for async consumers across the platform.

    Technologies

    Python, Flask, PostgreSQL, RabbitMQ, Docker, Kubernetes, AWS, Pytest, GitHub Actions

  • Jan 2023 - May 2023
    Co-Founder / CTO

    SuperAPI was a YC W23 company building a plug-and-play API cache with automatic CDC-driven invalidation. As co-founder and CTO, I spent three months in San Francisco for the batch and built a gateway that sat in front of customer APIs with no code changes, cached responses in under 20ms on hits, and used database change events to invalidate stale data instead of relying on TTL expiry alone.

    Key contributions

    • Co-founded SuperAPI (YC W23), a CDN API cache that refreshed responses from database CDC instead of TTL expiry, reducing stale reads and improving API response times for downstream clients.
    • Built the production stack on AWS with CloudFront, Lambda, DynamoDB, and Redis, using Debezium and Kafka for CDC-driven invalidation and in-place response updates.
    • Owned end-to-end architecture from CDC ingestion through CloudFront cache refresh; built observability for global latency and cache effectiveness.
    • Shipped SuperAPI as a plug-and-play API cache gateway with no SDK or code changes, serving cache hits in under 20ms with CDC-backed invalidation (trysuperapi.com).

    Technologies

    TypeScript, AWS, CloudFront, Lambda, DynamoDB, Redis, Debezium, Kafka, PostgreSQL