written & engineered by Arlen Frederick Kumar Berkeley, CA · updated June 2026
turn the page · tap a balloon to pop it↓
the big idea
The next great big buyer? It isn't a Who.
It's an agent that browses and buys things for you.
It compares and it picks, it decides and it pays —
while the humans have wandered away from their days of clicks.
But the web was built old, for the eyes and the thumbs,
so the brands get filtered before the deal comes.
So I build the translator — fresh, structured, and true —
so the old web can talk to the buyers brand-new.
where I showed up · 1
Co-Founder & CTO, Wrodium
2024 → now · Berkeley SkyDeck Batch 21 · 🏆 Most Innovative Technology
I co-built a company — GEO's the kind — keeping knowledge fresh and brands top-of-mind.
Closing a $1.5M pre-seed at a $15M post-money cap — Verdict Capital, 359 Capital, Harlem Capital & The House Fund.
Architected the whole product: Living Articles, the Semantic Resonance Engine, ClawdBot/OpenClaw, Wrodium Roast & the WebMCP layer.
Designed a revenue-weighted GEO framework — Conversion Potential Score, Prompt-to-Revenue Velocity, AI Traffic CAC — turning GEO into real dollar attribution across ChatGPT, Gemini, Perplexity & Claude.
Built GTM end-to-end: 17-inbox outbound on Instantly, Apollo, the SAFE workflow, Delaware filings & the Supabase → InsForge migration. Positioned vs Profound, Peec AI, AthenaHQ & Relixir in a market growing 34% a year.
where I showed up · 2
Researcher, Hearst Lab · UC Berkeley
2024 → now · advised by Prof. Marti Hearst · NLP & IR
In a lab full of language, with Hearst at the helm, I measure how robots make sense of the realm.
Co-authored GEO-16 — a 16-task benchmark for generative engine optimization. arXiv:2509.10762
Co-authored CHASE (submitted to COLM 2026) on knowledge freshness & citation behavior in retrieval-augmented LLMs.
Produced a Lean 4 formalization of GEO-16 for CS 294-268 — empirical NLP meets machine-checked proofs.
In talks with CalCompute (UC public compute, CA SB 53) on knowledge-freshness infra for auditable, legally-compliant AI.
where I showed up · 3
Founder, Air Quake Simulations
acquired · VR flight-sim hardware · founded at 19 ✈️
At nineteen I built, in a workshop of mine, an F-18 cockpit — and it flew just fine.
Founded & exited a VR flight-simulator hardware company building F-18 cockpit systems. Shipped hundreds of units before acquisition.
3D-printed cockpit parts that undercut the industry 10× — opening pro-grade sim hardware to a priced-out market.
Engineered an AI VR training platform in Python + PyTorch with reinforcement learning, real-time adaptive feedback & GPU pipelines in Unity3D.
the school part
B.A. CS, Data Science & Economics
UC Berkeley · Spring 2026 🎓
Computer Science, Data, and Econ — a fine three — at Berkeley in spring twenty-six, that'll be me.
Research on AI deepfakes as an epistemological break from historical deception, econometrics & ML theory.
on the mic
“The Ad-ification of AI: When Chatbots Become Salespeople”
CITRIS · UC Berkeley · Feb 2026 🎤
On ads inside chatbots — the trust we might lose — and the rules being drawn for the AI we use.
things I know how to do
Thing 1, Thing 2… Thing 8
From Python to proofs, from fresh crawls to schema — here's the whole truffula toolkit-arena.