Ashish B.

AI solutions architect, twenty years in. I design AI systems that survive production, cut what they cost on AWS and Google Cloud, and act as technical lead while you hire your first engineer. Clients hire me for the review before the build. YOU MIGHT BE HERE BECAUSE — Your AI feature works but the OpenAI, Anthropic or cloud bill keeps growing — You need an architecture review before you build, launch or raise — Your team is piloting AI and nothing is reaching production — You need a fractional CTO or technical advisor, not a full-time hire yet — You're a 50- to 500-person company that wants AI in its operations and doesn't know where to start — Your database or API got slow when you got customers — Your infrastructure was set up once and nobody understands it now — You want a self-hosted AI agent running your operations, on your own infrastructure, deployed properly WHAT I WORK WITH AWS, Google Cloud, Azure · Kubernetes, Docker, GitHub Actions · Vertex AI, Anthropic, OpenAI · Hermes Agent, Ollama · Postgres, MongoDB, Redis · FAISS, Milvus · Python THREE THINGS THAT DON'T FIT ON A CV — Architected the cloud AI/ML platform for BotSupply through to its acquisition — Kubernetes, zero-downtime model releases, 99% uptime under SLA for four years — Led cloud infrastructure and ML Ops for AirBills, a B2B payments platform, from launch to $6M ARR — designed for $1M in daily transactions — Two serial entrepreneurs have come back for their next project CLOUD AI WORK — Ops automation with self-hosted agents: deployed Hermes Agent for a 250-person company to automate operational workflows — 5.0, "deliverables matched the brief without hand-holding" — B2B FinTech: cut a consumer fintech platform's operating cost from $14,000 to $7,000 a month by re-architecting cloud workflows on Vertex AI and Anthropic — B2B Banking: policy drafting from 2–3 months to 1–2 weeks with AI-based cloud solutions, working with executives at Royal Bank of Canada — Drug approval intelligence: architected a cloud product for a client assisting drug discovery and predicting FDA approval changes, on AWS — AI content platform (in progress): AI-assisted content generation and scheduling application — the largest engagement on this profile — API & database performance: six-month engagement optimising a production API and database layer — 5.0, two contracts, same client — B2C News: cloud-native AI for summarisation, recommendation, image tagging and FAISS search on AWS and Azure HOW TO START Tell me what you're building or what's running, and what it costs. I'll tell you what I'd change first and why. If I'm not the right person, I'll say so.