Callum van den Enden

• Founded, designed, and shipped hey anna - "the analyst that tells you why your numbers moved." Connects Stripe, HubSpot, GA4, Mailchimp and 25+ tools; real statistical analysis (forecasts, significance tests, segmentation); themes verbatim text; writes exec-ready reports.
• Positioning: "analyst, not chatbot." Pricing sub-$1/day vs $400+/mo incumbents. 13 core features incl. AI Formulas =AI(), Smart Columns, AI Memory, Public Reports - solo, bootstrapped, zero-to-paid in under 6 months, live with billing.
• Stack: React + Cloudflare (Workers, D1, R2), Anthropic Claude primary, Paddle billing, PostHog.

User research SaaS, bootstrapped, ~50 staff, thousands of customers including Google, Figma, HubSpot, Asana, Zendesk, L'Oréal, Hilton, Airtable, Miro, Aesop, Trip.com, DuckDuckGo, British Red Cross, GoDaddy, Hexagon.

• Leads the Panel team (research-participant marketplace) and the research-analysis team building an in-product AI agent for results analysis.
• Prototyped an in-browser eye-tracking system (WebGazer.js gaze capture, I-VT fixation detection, AOI analysis, Redis/Sidekiq → PostgreSQL JSONB; privacy-first, no video persisted) - testing whether System-1 attention can be measured at consumer-webcam fidelity, replacing $30k lab hardware.

Pre-seed-stage enterprise AI content platform, team of 7, reported to CEO.

• Repositioned from $29 consumer to enterprise SaaS - killed consumer tier, set $999 minimum, rebuilt ICP/pricing around brand teams; scaled $5k MRR → $1M ARR in 22 months (3.3× YoY) across 48 brands incl. Sportsbet, the NBL, Australian Grand Prix, NBA; closed $100k contracts with Zscaler and the NBA.
• Shipped autonomous prompt optimisation (embedding-based RL across 15 style attributes, cosine-similarity convergence safeguard; 273+ experiments cut human edit rate 30%, zero-edit approval to 75%; authored R&D paper).
• Built an agentic workflow engine - reduced agent delivery time from 3 weeks to 2 hours.
• Shipped an ML video pipeline for the NBA/NBL (face/ball tracking, scene detection, auto-clipping, vertical auto-crop, multilingual subtitles in 6 languages) at 30× manual cadence.
• Stood up the analytics stack (AWS Glue ETL, DuckDB on R2, Athena, QuickSight). Acted as primary designer.

Custom Plates unit, $100M+ P&L, team of 9.

• Designed and prototyped an LLM-based pricing model interpreting the language of letter-number combinations as input to a deterministic pricing engine; now in state-wide rollout (6M+ drivers).

Growth-stage travel marketplace; teams up to 20 incl. 3 direct-report PMs. Reported to CEO/CPTO.

• Rescued Trip Planner - repositioned as an agent-led quoting tool, eliminating the cold-start problem; one-click checkout with commission attribution; shipped in a week; 30k WAU, +30% AOV.
• Shipped a recommendation engine across web/mobile/email (matrix factorisation/ALS, item-item & user-user CF, content-based embeddings, Thompson-sampling contextual bandits); 5% margin lift, 20% EDM revenue increase.
• Stood up behavioural analytics from scratch (Snowplow, PostHog, Heap).
• Drove 300% Flights revenue growth and led UK market entry $20M → $60M in 12 months; shipped NLP voice analytics on 100k+ call transcripts (+5% agent conversion).

• Managed teams of 4–8 across two stores; top 10% globally of 2,000+ stores.

• Applied mechanistic interpretability, optimising generative outputs toward target coordinates in ImageBind's 1024-dim shared embedding space; concept algebra across text/image/audio/video. PyTorch on Apple Silicon (MPS); ImageBind; SDXL VAE, AudioLDM 2, SVD; custom regularisers.

• Double major: Economics & Politics, Econometrics focus. Distinction.