Ali K.

I build the things that take repetitive work off your plate — workflows, internal tools, and Al systems that actually run in production, not just demos. Most of my work falls into two buckets: automating operations (data entry, reporting, syncing between tools, scraping, internal dashboards - the boring stuff that eats hours every week) and building Al into products (chatbots, support agents, document Q&A, anything that needs a model trained or fine-tuned on your specific data instead of a generic prompt). If either of those is the problem, that's the work I take on. On the technical side: Python and FastAPI are my main backend stack — REST APIs, async pipelines, background jobs, all of it. For Al work I do model fine-tuning, RAG pipelines (chunking, embeddings, vector search - Pinecone, pgvector, Weaviate, whatever fits the budget), and LLM integrations across OpenAl, Anthropic, and open-source models, including the routing and caching logic that keeps latency and cost sane at scale. I'm backend-first, but with Al-assisted tooling I also handle frontend (React/Next.js) when a project needs one person to own the whole stack instead of two. For deployment I work across AWS and Azure (S3/Blob, Lambda, EC2), ship frontends to Vercel, and run backends on Render or Railway, with Redis for caching and Celery/RQ for anything that needs to run async. Before this I ran my own service agency for a few years, so l'm used to owning a project end to end, not just writing code and disappearing. If something's a bad idea or wrong scope, I'll say so before you spend money on it. Send me what you're trying to build or automate and I'll tell you straight away whether it's feasible and roughly how long it'll take. RAG, Agents, n8n, make, CRM, Hubspot, Elevenlabs, voice agents, chatbot