Internal automation: Scripts that sync data between systems, clean messy spreadsheets, or generate reports without manual copy-paste.
LLM-powered tools: RAG systems for internal knowledge bases, chatbots that pull from your docs, or AI assistants that automate repetitive tasks.
Data pipelines: Extractors that pull data from APIs, transform messy inputs into structured outputs, and load results where your team needs them.
Scripts that automate manual workflows: data extraction from APIs, file processing pipelines, report generation that runs on schedule. Error handling that logs failures instead of silently breaking.
I build tools that work reliably in production, not just demo scripts that break when inputs change. Proper logging, validation checks, and documentation so your team can maintain them.
Retrieval-Augmented Generation (RAG) systems that let LLMs answer questions using your internal docs, support tickets, or knowledge bases. API integration with OpenAI, Anthropic, or local models.
I handle chunking, embeddings, vector stores, and retrieval logic so your team gets accurate answers from AI without hallucinations or "I don't know" responses when the info exists.
Connect systems that don't talk: pull data from CRMs, marketing platforms, or analytics tools, transform it, and push to databases or dashboards.
I build pipelines that run reliably, handle rate limits and API errors gracefully, and log failures so you know when something breaks instead of discovering it weeks later.
Custom web apps for your team: admin panels, data entry forms, approval workflows, or reporting dashboards that pull live data.
Built with Flask, FastAPI, or Streamlit depending on complexity. Deployed on your infrastructure or cloud platforms. Designed for internal use, not public-facing polish.
AI assistants that integrate with your data: customer support bots that pull from your docs, internal search tools that understand natural language queries, or agents that automate multi-step workflows.
I handle prompt engineering, context management, and tool calling so the assistant actually completes tasks instead of just generating text.
Scripts documented so your team can modify them. Error handling that logs issues and sends alerts. Version control in GitHub so changes are tracked.
I build for maintainability, not just "works once." Your team should be able to understand and extend what I build after I'm gone.
I specialize in AI-powered automation and LLM integration for marketing and operations teams. Most developers build product features. I build internal tools, scripts, and AI assistants that automate workflows your team does manually. Faster iteration, lower cost than full-time eng.
Simple scripts (data extraction, report generation) take 1-2 weeks. API integration pipelines take 2-4 weeks. RAG systems or LLM-powered tools take 4-6 weeks depending on data complexity and feature requirements. I provide working MVPs fast, then iterate.
I can deploy to cloud platforms (Heroku, Railway, Render) or set up basic infrastructure. For complex deployments, I work with your eng team or recommend contractors. Most internal tools don't need enterprise infrastructure — they just need to work reliably.
Yes — Claude, Cursor, GitHub Copilot. But AI-generated code still requires human review, testing, error handling, and architecture decisions. I use AI to code faster, not to skip the engineering work that makes tools production-ready.
You get: working code in GitHub, documentation on how it works, deployment instructions, and a handoff session with your team. I can provide ongoing support, but most tools are designed to run without constant maintenance. If something breaks, I'm available to fix it.
Python (FastAPI, Flask, Streamlit), JavaScript when needed, SQL for data, cloud platforms for deployment. For LLM integration: OpenAI, Anthropic APIs, LangChain, vector databases (Chroma, Pinecone). I choose based on your requirements, not what's trendy.
If your team is doing manual work that could be automated, or you want AI-powered tools but don't have eng resources, let's talk.
Book a Strategy Call →