Senior AI engineer, 10+ years in software. I take your idea from napkin sketch to deployed product — chatbots, RAG pipelines, AI agents, the works. You get something real your users can touch, not a pitch deck.
What I build
Autonomous agents that handle multi-step workflows. I've built agents for everything from data pipelines to customer support triage.
Chat-with-your-data systems. Upload documents, ask questions, get real answers with source citations. Built on vector search and LLMs.
Drop Claude, GPT-4, or Gemini into your existing product. I handle prompt engineering, caching, rate limits, and cost controls.
Customer-facing conversational AI. Not the kind that says "I don't understand" every other message — ones that actually work.
Let non-technical users query databases in plain English. I built one of these for a healthcare company's medical billing system.
End-to-end builds: backend API, frontend UI, LLM layer, deployment. One person, one codebase, no coordination overhead.
How it works
We talk for 30 minutes. You tell me what you're building and who it's for. I tell you if I can help, what it'll take, and what it'll cost. No pitch, just a straight conversation.
I build your product in 1–4 week sprints with async updates every day or two. You see real progress, not status reports. If something needs to change mid-build, we adjust.
You get a deployed product, clean code, documentation, and a handoff call. I stick around for two weeks after launch to squash any bugs. Your product, your repo, no lock-in.
Things I've built
Built an AI agent at Podium that runs full sales conversations for auto dealerships — from first contact to close — without a human in the loop. Also built a Customer Data Platform that pulls from multiple DMS and CRM integrations so the agent has real context (lead car equity, history) for every conversation. The system runs on Podium's platform serving 100,000+ local businesses.
Built the core AI pipeline at Caddi. A Chrome extension records how users do their work (browser clicks, DOM changes, network events), then a video processing pipeline extracts the key frames and combines them with transcriptions and LLMs to generate structured workflow representations. Users can query and modify their automations through a natural language chatbot powered by Claude. Also built an internal graph editor (like Zapier) to manually fix automation loops when the AI output is incomplete.
Built a pipeline at Revado AI that lets billing specialists query medical codes in plain English instead of writing SQL. Cut manual lookup effort by 40%. Also built a claims dashboard for denied claims, an automated denial-and-appeal workflow using speech recognition, and a PyTorch model that predicts claim rejections before they happen.
About
I'm Devin Jackson. I've been writing software for over a decade, and for the last two years I've been focused on AI engineering full-time.
Right now I'm a Senior AI Engineer at Podium. Before that I was at Caddi (AI engineering), Revado.AI (where I built a Text-to-SQL pipeline for medical billing), and Nickels (where I led development of a generative AI customer tool that drove a 20% engagement lift). Before the AI work, I spent 3.5 years as a senior engineer at LeafLink building a B2B e-commerce platform.
I started this agency because I kept meeting founders who had solid AI ideas but no one to build them. Agencies wanted $50K+ and months of timeline. I can usually ship something real in 2–4 weeks for a fraction of that cost.
I work remotely from Maryland. Most of my clients are US-based startups between pre-seed and Series A.
Book a free 30-minute call. I'll tell you straight whether I can help and what it would take.