Senior LLM Engineer & Odds API Integration Specialist (Claude, The Odds API, Supabase)

We’re building Big Mike — a sports-betting AI that delivers +EV picks and live odds reads to users over iMessage/SMS, like a sharp friend texting you the line, not a generic chatbot. The product is on TestFlight: iOS app, Supabase backend, messaging flows, and a picks pipeline that ingests qualifying bets, stores them, and broadcasts them in Big Mike’s voice. We need a senior contractor to own the odds data layer, pick delivery system, and LLM integrations end to end. Picks are the product. If you don’t have real experience with sports odds APIs, line movement, EV modeling, or pick distribution systems, this role is not a fit. What You’ll Work On Odds & Picks (Primary Focus) Integrate and maintain The Odds API — live spreads, totals, moneylines across US books (Pinnacle, DraftKings, FanDuel, etc.) Build and improve odds polling, caching, and line-movement detection (open vs. current line, sharp signals, notable moves) Own the pick ingestion pipeline — webhook receivers (e.g. OddsSoul), idempotent storage, EV/confidence filtering, and user targeting by sport/team preferences Work with pick data fields in production: offered odds, fair odds, EV%, Kelly units, confidence tiers, sharp signals, competing books, line at send Ensure picks flow correctly from odds signal → database → Claude copy → SMS/iMessage delivery → result tracking Improve pick qualification logic — what counts as “strong” vs. “solid,” when not to send, and how to surface the best number across books Support line-movement and slate alerts driven by live odds changes, not just static cron messages LLM / Prompt Engineering Maintain and extend our Claude (Anthropic) system for a human iMessage persona — short, direct, no bot voice Inject live odds context into inbound Q&A (current lines, book differences, movement) so answers are grounded in real numbers Generate proactive pick SMS from structured pick payloads (bet, line, book, EV, sharp note) with strict voice rules Optimize for latency, cost, and quality — prompt caching, context assembly from DB caches, model selection Handle edge cases honestly: no pick tonight, thin read, line already moved, user asks about a bad number API Integration & Backend Extend Supabase Edge Functions (TypeScript/Deno) that orchestrate odds fetches, pick processing, LLM calls, and outbound messaging Integrate messaging providers (Linq, Twilio, BlueBubbles) with delivery tracking and failure handling Wire webhook receivers with validation, deduplication, secrets, and production logging Connect injury/schedule context (ESPN, Sleeper, etc.) where it affects pick relevance and alerts Write observable, maintainable code our team can extend after handoff Required Experience Must have: 3+ years shipping production systems that touch sports betting odds data Hands-on experience with The Odds API or comparable odds feeds (Pinnacle, Sportradar, OddsJam, OddsAPI alternatives) Understanding of American odds, spreads, totals, juice, line shopping, sharp vs. public books, and line movement Experience building pick/alert pipelines — ingest, qualify, store, distribute, and track results 3+ years with LLM APIs in production (Anthropic Claude strongly preferred) Strong TypeScript and REST/webhook integration (serverless/edge: Supabase, Lambda, Workers, etc.) Track record with SMS/iMessage or conversational alert products Clear English communication and ability to work inside an existing codebase Strong plus: +EV / expected value modeling or working with pick engines that output fair odds and EV% OddsSoul or similar pick-feed webhook integrations Supabase (Postgres, Edge Functions, cache tables, cron jobs) Push notifications and iOS backend hooks Prompt evals for sports copy — making picks sound human without hype or guarantees Example Deliverables (First 2–4 Weeks) Audit the odds + picks stack — The Odds API usage, line-movement cache, pick webhook flow, and where data goes stale or drops Harden pick ingestion — idempotency, validation, error handling, and logging for the full pick → SMS path Improve one odds-driven feature — e.g. better line-movement alerts, sharper live odds context in inbound replies, or pick targeting by book/line quality Technical doc — how odds data, picks, and Claude context connect; how to add a new market or sport safely Engagement Details Type: Contract / freelance (long-term possible) Hours: 20–40 hrs/week Duration: 1–2 weeks initial; extension likely How to Apply Generic cover letters will be ignored. Include: 2–3 examples of odds API or pick/alert systems you built (links or short case studies — redact secrets) One paragraph on how you’d design a pipeline from live odds → qualifying pick → personalized SMS alert Which odds APIs you’ve used and what data you pulled (markets, books, movement, etc.) Answer: How do you decide a line move is “notable” vs. noise? Answer: How do you prevent the same pick webhook from sending duplicate texts? Confirm comfort with TypeScript, Claude Messages API, and The Odds API What We’re Not Looking For Chatbot builders with no sports odds background “I used ChatGPT for content” — we need production API + odds domain depth Prompt-only freelancers who can’t wire webhooks and databases Agencies that bait-and-switch to junior devs About Us Small, fast team building a consumer product at the intersection of sports betting, live odds, AI, and messaging. Picks and line quality are what users pay for. You’ll ship to real users and see impact immediately. Please type penguin in your letter cover.