Can a Senior React Engineer Also Build Real AI Features?

Why "both sides" is the whole point

AI features fail at the seam between the model and the interface wrapped around it. Streaming tokens into a UI, cancelling an in-flight agent run, rendering tool calls as they resolve, degrading gracefully when a provider is slow — these are frontend problems and backend problems at the same time. When the React engineer and the AI engineer are two different people, every one of those becomes a ticket, a meeting, and a week of latency.

Put both roles in one head and the negotiation disappears. The person building the model gateway is the same person building the component that consumes it, so the contract between them is designed once and iterated in hours. That is the difference between an AI feature that ships this month and one that stalls in integration.

Frontend depth: real interfaces, real deadlines

Senior React work is not measured in tutorials completed; it is measured in interfaces that survived contact with real users and hard dates. Blake's are specific and named:

• Snap! Spend — owned all commits on the customer-facing React app at a neobank, the surface real customers used to move money.
• The Emmys — built a WebRTC platform in React in five weeks, a live-event bar where there is no second chance to ship late.
• Medaxion — a React dashboard on top of a Rails EHR, which means healthcare-grade data on a production frontend, not a toy.

That is ~a decade of React and TypeScript behind interfaces where correctness and uptime actually mattered.

AI backend depth: the parts that keep AI alive

The easy 80% of an AI feature is wiring an API key to a chat box. The hard 20% is everything that keeps it working after the demo. Blake builds that layer directly:

• Model gateways that route tool-using agents, enforce policy, and fall back gracefully when a provider degrades.
• RAG pipelines wired to real corpora, with retrieval quality measured rather than assumed.
• Evaluation harnesses that backtest output against historical ground truth, so "better" is proven, not vibed.
• A live real-time voice agent you can pick up and talk to — a far harder bar than a text demo, because latency and interruption handling are unforgiving.

Full-stack payments, proven at scale

The clearest evidence that one engineer can carry a whole system is a load-bearing platform where money moved and downtime was not an option.

As Director of Engineering at a neobank for roughly three years, Blake led 17 engineers across 5 products and ran $720M in payments at 100% uptime — Stripe Connect multi-tenant, Plaid ACH, Apple and Google Pay, fraud rules, and KYC/KYB. He owned all commits on the Snap! Spend customer-facing React app and wrote core payments code across the platform. Separately, he built a multi-tenant payments + field-ops platform on Stripe, Temporal, and Postgres in Rust/Node on GCP — the same multi-tenant, workflow-driven shape that production AI features live inside.

Delivery speed you can point at

Range and depth mean nothing without pace. Two concrete data points: a WebRTC platform for the Emmys built in React in five weeks, and a React dashboard delivered on a Rails EHR (Medaxion). Beyond client work, Blake built Nectar, a programming language in Rust that compiles to WebAssembly with a public compiler and 2,500+ tests — the kind of systems artifact you cannot fake — and a tamper-evident audit trail signed with post-quantum cryptography (ML-DSA), with a provisional patent filed. The positioning across all of it: production AI whose every decision is signed, anchored, and independently auditable.

Frequently asked questions