Senior AI Engineer, Agent Platform

Company Description Intellectus Partners, LLC is a modern wealth advisory firm built by and for innovators, entrepreneurs, and business owners. The firm focuses on tightly aligning clients’ business interests with their personal wealth strategies, rather than treating them as separate worlds. Intellectus builds AI systems for Family Office, UHNW and institutional client cohorts. We are hiring an Applied AI Engineer to own a major surface of the platform end-to-end.Our Dev team is a talented, hard working, small, senior team shipping production software to real users with real capital. We work fast, we ship, and we hold ourselves to an institutional bar. can expect to join a team that values analytical thinking, innovation, and close collaboration with clients who are building and scaling businesses. The environment emphasizes both technical excellence and forward-looking financial insight.Role Description - Take existing product from late state development to hardened and shipped. The Senior Applied AI Engineer, Agent Platform is a full-time(could start Part-time) role based in the San Francisco Bay Area, with flexibility for partial work from home. This role is responsible for designing, building, and maintaining AI-driven agent platforms that support internal teams and enhance client-facing wealth management solutions. Day-to-day activities include developing and integrating models for natural language understanding, pattern recognition, and decision support; architecting and optimizing neural network–based components; and collaborating with product, advisory, and engineering stakeholders to translate business requirements into scalable AI services. You'll own a meaningful slice of the engineering surface. The work spans four areas:- Agent systems — building features powered by foundation models that operate inside real workflows, not as chatbots bolted on the side.- Data and memory — the discipline of what the system knows, where it came from, and how long it stays true.- Safety and supervision — the layer that decides what an AI feature is allowed to do, and what it has to escalate.- Quality — making sure model swaps, prompt changes, and new releases never silently regress.Production discipline — security posture, monitoring, audit readiness, compliance evidence — is a property of how you build, not a separate track. We expect every feature to ship that way.---The stack familyFront end — Next.js 15, TypeScript, React, TailwindBackend — Supabase (Postgres, Auth, RLS, Storage) with field-level encryption and end-to-end audit loggingEdge runtime — Supabase Edge Functions on Deno / TypeScriptAI layer — Claude (Anthropic SDK) as the primary surface, with a model router for provider fallback and cost-tier routing across the major frontier providersIteration loop — GitHub as source of truth, Lovable for the front-end iteration loopCI / observability — GitHub Actions, Sentry, PostHogSecurity posture — SOC 2 readiness in progress, SSO + MFA, pentest cycle on the calendarIf those words mean something to you and you have shipped on this kind of stack in production, we should talk.You should apply if:You have shipped AI in production. Not prototypes, not demos. Multi-step tool-calling features that real users hit, that survived the second month after launch. You have debugged what happens when a model returns malformed JSON at 2 AM under load.You are fluent with LLM APIs at depth. Tool use, structured-output validation (Zod or JSON Schema against model responses), retrieval, streaming with mid-stream error recovery, token-budget and context-window discipline, and cost-tier routing. You have opinions about when to use a smaller model and when not to.You are production-grade on the stack. TypeScript and React (Next.js app router, server components, server actions). Postgres at the query level, not just through an ORM — you can write RLS policies, SQL functions, and triggers. Supabase or equivalent BaaS experience. Comfortable in a Deno / Edge Function runtime, or fast enough to be productive in one within a week.You understand production security and compliance for real. PII handling, field-level encryption, RLS design that actually holds up, secrets discipline, audit trails that survive examination. You think in OWASP top-10 and STRIDE without being asked to.You have shipped observability for AI features. Token costs, latency, success rate, prompt versioning, regression detection. You know what "the model got worse this week" looks like in a dashboard.You can hustle. Early-stage platform, real users, real money flowing through it. We move fast, we ship, we own what we ship.Nice to havePrior work on agent systems specifically — tool registries, multi-step planning, supervision wrappers.You have been through a SOC 2 audit on a system you helped build, not just observed one.Experience integrating with financial data APIs at the protocol level — REST plus webhook reconciliation, idempotency, replay safety.You have built or worked with a model router across providers.You know finance well enough that you do not need a translator.---How we workYou will work directly with the Founder/CEO every day. Async-friendly, decision-fast, no committee. The platform is the deliverable. The team is small and senior and expects you to be the senior-most engineer on the surface you own.This is a long-term engagement. 20–40 hours per week to start, scaling to full-time as the platform load grows. Open to a contract-to-hire structure for the right person.We will not discuss architecture details, product strategy, or specific integrations in a public posting. Those conversations happen under NDA, once we know we're talking to the right person.---To applySend a short email — jared@intellect.us1. A production AI feature you shipped. What it does, the stack, the model, and what you learned the hard way after it went live.2. A time you had to choose how much autonomy to give an AI feature — whether it should suggest, draft, act with confirmation, or act on its own — and what made you choose. If you have never made that call explicitly, tell us how you would think about it.3. Your production security posture. A real example of how you have handled PII, encryption, access control, or audit evidence in a system you built.4. Availability and engagement structure you are looking for.Apply to jared@intellect.us with the subject line **Applied AI Engineer — [your name]**.