I build the systems behind AI products.
Founding engineer, fractional CTO, or a senior/staff backend role. NYC or remote.
I'm CTO at LoanBlocks, where I own the backend end to end — document ingestion, extraction pipelines, and LLM orchestration for institutional lenders.
Three ways to work with me
Different commitments, different stages. Here is what you get in the first month of each.
Founding engineer
Pre-seed to Series A, first or second engineer
- Stand up the backend, the deploy pipeline, and the on-call story
- Own the data model and the API surface the next engineers build against
- Ship the first production version with you, then help hire the team around it
Fractional CTO
Founders who need technical direction without a full-time hire
- Architecture review and a roadmap you can hand to a contractor or a new hire
- Engineering hiring — screening, interview loops, leveling
- Infrastructure and vendor decisions, including what they cost at your stage
Senior / staff software engineer
Series A onward with an existing team. NYC or remote
- Backend services in Python and TypeScript on AWS
- LLM integration that survives a provider outage — retries, backoff, failover
- High-availability APIs and the ingestion pipelines behind them
About
Where I've been, and how I work.
- Built the LoanBlocks backend end to end — document ingestion, extraction, and LLM orchestration on AWS, in Python and TypeScript.
- Designed the PostgreSQL and Prisma schema for Productivity Sumo. Query performance improved 30%.
- Shipped a Next.js and Go dashboard handling 1M+ daily transactions. Caching took queries from 2s to 200ms.
- Built an ESG score analyzer on Next.js and FastAPI over 100k+ daily environmental metrics. Won most innovative at a climate hackathon.
I've been the first engineer twice, which means I've also been the person carrying the pager. That shapes what I build: LLM calls get retries and a second provider because one provider will go down, and I'd rather design the schema once than migrate it under load.
Startup Experience
Productivity Sumo
Founder and Software Engineer
Productivity Management Software
LoanBlocks
CTO, Co-Founder, Founding Software Engineer
Derisk Credit Agreements with AI Covenant Review
SignCode Inc
Founding Engineer
Municipal Sign Code Ordinance Checks for Sign Makers
RiskHelm
Founding Engineer
First-Party Insurance for AI That Acts on Its Own
Validus Nexus
Software Engineer
AI Business Matchmaking for Founders, Funders, and Family Offices
Immpression
Founding Engineer
Mobile Marketplace for Buying and Selling Art
What I work with
Backend and systems first. The frontend is real, it just isn't the reason to hire me.
Where I've used it
LLM orchestration
Multi-provider failover, retries with exponential backoff, and priority-based task routing — at LoanBlocks and in Fleet.
Document pipelines
Ingestion and extraction for credit agreements and covenant review, running on AWS for institutional lenders.
Throughput
A Next.js and Go dashboard handling 1M+ daily transactions with WebSocket streaming.
Query performance
Caching took dashboard queries from 2s to 200ms. Schema design in PostgreSQL and Prisma cut Productivity Sumo query time 30%.
Scale on FastAPI
An ESG score analyzer processing 100k+ daily environmental metrics with sub-second responses.
Zero to production
Productivity Sumo built from scratch — schema, Stripe billing, and Google OAuth. Immpression shipped on React Native over Node, Express, and MongoDB.
Languages
Python
TypeScript
Go
JavaScript
Backend and APIs
Node.js
Express.js
Django
Flask
GraphQL
- FastAPI
- WebSockets
- REST
AI and LLM systems
LLM orchestration
Multi-provider failover
Retries and exponential backoff
Document ingestion and extraction
Priority-based task routing
AWS and infrastructure
AWS
Docker
Kubernetes
Vercel
- AWS Certified Cloud Practitioner
Data
SQL
MongoDB
- PostgreSQL
- Prisma
Frontend
React
Next.js
Tailwind CSS
Redux
React Native
React Query
Tools
Git
GitHub
Jest
React Testing Library
Figma
- Linear
Book an intro call
Pick a time that works, then a couple of quick questions so I can come prepared. Fifteen minutes.