Igor G.

I help SaaS teams turn unreliable AI/data systems into measurable production workflows: Agentic RAG, ML evaluation, data pipelines, dashboards, and Stripe-backed revenue systems. Best fit: funded teams with a broken RAG chatbot, hallucinating AI workflow, weak ML evaluation loop, unreliable data pipeline, or payment/data reconciliation problem that needs a scoped production fix. Recent paid Upwork proof: Hilltown Media Group funded fixed-price work for a Pretix + Stripe Connect plugin. I shipped destination-charge payment flow work, refund-fee-shield behavior, test coverage, and handoff evidence across M1 + M2. That proof matters because I do not start unfunded work and I ship code/tests, not slideware. Agentic RAG / ML work I am strongest at: - RAG evaluation harnesses: retrieval quality, grounding checks, failure cases, regression tests - Agentic workflow hardening: tool-call gates, budget limits, trace review, rollback paths - Data Science / ML diagnostics: forecasting/model sanity checks, feature pipelines, dashboard QA - Production data systems: Python, SQL, TypeScript, BigQuery/GCP, AWS, APIs, observability - SaaS revenue systems: Stripe Connect, webhooks, payout/reconciliation logic, marketplace flows Typical fixed-scope starting points: - $500-$750 AI/data audit with prioritized fixes - $750-$1,500 RAG or ML evaluation harness - $1,000-$2,500 data pipeline, dashboard, or Stripe/data reconciliation slice I am not a good fit for unfunded prototypes, vague "build my whole AI app" requests, or hourly work without a funded milestone. If the scope is real, I will make the risk visible, define the smallest escrow-funded slice, and ship the evidence.