Full Stack: TypeScript/React, Python, Node, PostgreSQL

Job Description

Job Description

How to apply

1. Apply through this posting only. Attach a resume, and in the space provided include one of the following:
   

• A link to something you built that is live
• A GitHub profile with commits we can read
• A short note on the hardest integration or AI problem you've solved, and what broke
  

Applications with none of the three won't get far. One real thing beats five pages of skills.

We're a technology consultancy in Dallas that builds and runs the systems our clients depend on — including the software that doesn't exist off the shelf. We've been doing this a while, and the work is real.

This role centers on growing software footprint. There's software already in production to own and extend, several AI projects in flight, a new client-facing portal to build, and a steady supply of problems nobody has scoped yet. Most of your time lives there; some of it won't.

You'd own the architecture across all of it — some yours to design, some inherited and in need of a plan, and a fair amount with no established pattern to copy.

Read this before you apply

Software already running. Systems in production that people depend on daily. You'd own them — extend, modernize, and decide what's worth rebuilding versus leaving alone. Not everything here is greenfield, and knowing the difference matters.

AI systems, several in flight. Retrieval over the firm's documents, subject similarity search, agent workflows. Some exists and needs to get better; some hasn't been built. Most of it has no best practice to look up — we'd rather you prototype three approaches and measure them than ship the first one that works. That means evaluation sets, error analysis, and honesty about when a result is noise. "mostly right" is a defect.

Integrations. document storage, billing, intake, and the portal all have to agree with each other. REST, webhooks, OAuth, scheduled sync, and the occasional vendor whose API was clearly an afterthought.

Whatever we haven't scoped yet. The part we're most serious about. A firm our size generates more solvable problems than anyone has had time to write down. Part of the job is finding them and making the case for fixing them. If you need a groomed backlog, this will be frustrating. If you've been the person who noticed the problem and then fixed it, this is the job.

Required

• Shipped and maintained a production web application end to end — auth, data model, front end, deploys, and whatever happened after it went live
• Worked inside code you didn't write and made it better without breaking it
• Integrated systems you didn't build — REST APIs, webhooks, OAuth, data sync
• Built something on an LLM API that real users touched, with error handling and a plan for when the model is wrong
• Solved problems that had no established approach — surveyed the options, built more than one, measured which won
• Comfortable owning architecture decisions and working directly with end users and administrators — there's no product manager in between
  

Helpful

• Legal, healthcare, financial services, or another regulated industry — where confidentiality was a design constraint, not a policy PDF
• Data science or AI/ML research background: evaluation sets, error analysis, fine-tuning, embedding models, retrieval methods
• RAG or agent frameworks running in production
• Microsoft 365 and Entra ID identity, guest access, conditional access
• Experience as the only engineer on something that mattered
  

Stack

Newer work leans TypeScript/React, Python, Node, and PostgreSQL on Azure, with Microsoft 365 and Entra ID for identity. On the AI side: LLM APIs, retrieval, embedding models, vector search, evaluation.

Existing systems vary — deciding what stays and what gets rebuilt is part of the job. We're not dogmatic about tools; we care that you can explain a choice, not that you already agree with ours.

Why this job is worth your time

• Range, not a ticket queue. Legacy systems to improve, AI to architect, new software to build, and open ground to define.
• Room to invent. Several of these problems have no known answer. You'd research, try things, measure them, and be wrong occasionally — with the mandate to do that properly.
• A direct line. You'll work with our technical leadership and the people using what you build. No product committee, no eight-week discovery phase.
• You set the standard. You'd be our engineer on this work. The patterns and the bar for what "done" means start with you.
  

Arrangement

This is a contract-to-hire engagement. We start with a paid, scoped project so both sides can see how the work actually goes, move to an ongoing contract, and convert to full-time when it's clearly working. If you'd rather stay contract long-term, that's a conversation we're open to.

Rate and salary depend on experience and scope — tell us what you're targeting and we'll tell you quickly whether we're in the same range.

Hybrid, Dallas–Fort Worth. Some on-site time.

We're a technology consultancy in Dallas that builds and runs the systems our clients depend on — including the software that doesn't exist off the shelf. We've been doing this a while, and the work is real.

This role centers on growing software footprint. There's software already in production to own and extend, several AI projects in flight, a new client-facing portal to build, and a steady supply of problems nobody has scoped yet. Most of your time lives there; some of it won't.

You'd own the architecture across all of it — some yours to design, some inherited and in need of a plan, and a fair amount with no established pattern to copy.