Company: Streak Promoter
Location: Remote (US-friendly time zones)
Comp: Equity only at this stage (founder-level; standard 4-year vesting with 1-year cliff). No salary until we raise.
Streak is a new kind of event platform: ticketing + a promoter rewards marketplace + local business partnerships. Think mycelium network for IRL culture—connecting attendees, promoters, venues, and neighborhood businesses so everyone grows together.
We’re pre-seed and pre-revenue. You’d join as a founding team member and shape the product, data, and ML stack from day one. We are currently in the final stages of finishing the web beta and have venues and businesses ready to onboard and are ready to start implementing AI-powered features.
The Opportunity:

Competitors sell tickets. We’re building intelligence:

• Event & reward recommendations that feel like Spotify for going out.
• A living mycelium graph (users ↔ promoters ↔ venues ↔ businesses).
• Content assistants for promoters and venues.
• Dynamic reward matching and predictive ticketing/attendance.
• Your work will directly move core business metrics: discovery, conversion, repeat attendance, and partner ROI.

What You’ll Do

• Own recommendations end-to-end: telemetry → features → recall/rank/rerank → online serving → A/B measurement.
• Design the data contracts for events, rewards, promoter referrals, and conversions; partner with backend to instrument logging.
• Stand up the Feature Store (MVP) and Vector Index for semantic search/recs.
• Model the “mycelium” graph: build/maintain a graph of users/promoters/venues/businesses; ship graph-powered recall & similarity.
• Ship AI content tools: promoter copy, event descriptions; safe prompt/response pipelines with guardrails.
• Predictive models: initial ticket sales/attendance forecasts and triggers for dynamic promos.
• Productionize ML: AB tests, monitoring, drift checks, cost/perf tuning, incident playbooks.
• Collaborate deeply with Product, Design, and our NestJS team to ship fast behind feature flags.

Our Stack (you don’t need all of this—just be eager to own and learn)

• App/backend: NestJS (TypeScript), PostgreSQL, Stripe, AWS (S3, ECS).
• Data/ML: Python, dbt, Airflow, S3/Parquet + Warehouse, pgvector/Milvus, Neo4j/Neptune, Feast (or equivalent), ANN search, XGBoost/Prophet.
• LLM/RAG: Open-source + hosted LLMs, retrieval over events/rewards/FAQs.
• Ops: OpenTelemetry, MLflow, Evidently, feature flags/A-B (e.g., GrowthBook/ConfigCat/LaunchDarkly).

What Great Looks Like (Requirements)

• 4–8+ years building production ML systems, including at least one shipped recommender that moved a business metric.
• Strong Python + practical ML (feature engineering, AB testing, offline/online evals).
• Experience with ANN/vector search and at least basic graph modeling (Neo4j/Graph embeddings/GNNs or equivalent).
• Comfortable across data plumbing (streaming/ETL/dbt) and serving (APIs, latency, caching, fallbacks).
• Obsessed with measurement (metrics, guardrails, experiment design) and reliability in prod.
• Startup-ready: you prototype fast, instrument carefully, and ship.

Nice to Have

• LLM experience for assistive tools (prompt pipelines, safety/PII filtering).
• Forecasting/dynamic pricing exposure.
• AWS data stack (Kinesis/MSK, Redshift) and Feast.
• TypeScript/NestJS familiarity to collaborate closely with backend.

Day-0 to Day-90: What You’ll Ship

• Days 0–30:
• Telemetry schema + dbt models for events/rewards/promoter referrals.
• Feature Store MVP + pgvector set up.
• Content Assistant MVP (promoter copy & event descriptions) behind a flag.
• Days 31–60:
• Event/Reward Recs v1 (hybrid rules + embeddings) with online AB.
• Promoter Insights v0 (simple attribution & actionable suggestions).
• Baseline guardrails: PII scrubbing, prompt linting, abuse heuristics.
• Days 61–90:
• Dynamic reward matching (semantic + context).
• Forecasting MVP for ticket sales/attendance and promo triggers.
• Post-launch iteration plan based on experiment readouts.

Why Join Streak

• Founder-level ownership and a chance to architect the intelligence layer of a new marketplace.
• Hard, interesting problems at the intersection of recs, graph, and local commerce.
• Ship fast, learn fast, and see your work live in the world—nights out, packed rooms, real community impact.

Compensation

• Equity only until we raise (founder-level; exact range based on fit/scope).
• Standard 4-year vesting with 1-year cliff.
• We’ll formalize equity in written agreements and comply with applicable laws in your location.

Job Types: Full-time, Part-time
Pay: $22,000.00 - $100,000.00 per year
Expected hours: 20 – 40 per week
Work Location: Remot

• Own recommendations end-to-end: telemetry → features → recall/rank/rerank → online serving → A/B measurement.
• Design the data contracts for events, rewards, promoter referrals, and conversions; partner with backend to instrument logging.
• Stand up the Feature Store (MVP) and Vector Index for semantic search/recs.
• Model the “mycelium” graph: build/maintain a graph of users/promoters/venues/businesses; ship graph-powered recall & similarity.
• Ship AI content tools: promoter copy, event descriptions; safe prompt/response pipelines with guardrails.
• Predictive models: initial ticket sales/attendance forecasts and triggers for dynamic promos.
• Productionize ML: AB tests, monitoring, drift checks, cost/perf tuning, incident playbooks.
• Collaborate deeply with Product, Design, and our NestJS team to ship fast behind feature flags