Aavesh
Naykodi

I build AI systems that actually work.
And I write about the ones that don't.

1.6M+ Weekly active users
15+ Years in product
2x Ventures founded
Read Edge of Chaos → See the work
About

The kind of PM who
can read your pipeline spec

I build AI-native platforms end to end — strategy, org design, engineering SOWs, data pipelines, and the guardrail systems nobody else wants to own. Streaming, real estate tech, SaaS — the common thread is turning cross-functional chaos into systems that hold at scale.

At Classical California I rebuilt a legacy public radio org into a 1.6M+ weekly active user digital platform — a hybrid recommendation engine, an LLM-driven chat moderation layer, and a data activation framework that drove 84,800 new users in under 3 months.

Before that: computer vision at the MLS, 80% moderation overhead cut, an industry award. Co-founded two ventures — one got into Web Summit Alpha in Dublin. Three USC degrees (CS, Engineering Management, MBA). Zero interest in roadmaps that don't ship.

Current roleDirector of Product & Engineering, Classical California
TargetingStreaming, Marketplaces, AI/ML Products
Stack depthGenAI Agents, RAG, Vertex AI, BigQuery, Python, Elasticsearch
WritingEdge of Chaos — what PMs won't say out loud
Outside workAquascaping, planted tanks, aquatic ecosystems
LocationLos Angeles, CA
Substack

Edge of Chaos

All essays →
Edge of Chaos
AI product strategy, real case studies, and what actually ships vs. what gets demoed.
Subscribe →
Experience

What I've actually shipped

Sept 2020
Present
Director of Product and Engineering
Classical California (KUSC + KDFC)
Public media network · 1.6M+ Weekly Active Users · Los Angeles, CA
Unified two legacy broadcast brands into a single digital platform across web, mobile, and ambient devices. Built and led a 10-person cross-functional org. Shipped a 0-to-1 hybrid recommendation engine, LLM moderation layer for live chat, Knowledge Graph ingestion pipeline with Human-in-the-Loop review, and a unified CRM-AdTech-Streaming data activation framework. Cut feature cycle time 30%.
84.8K
New customers in 3 months
42%
YoY signup growth
2.1x
Day-7 retention lift
194%
More return visits per chat user
GenAI AgentsRAGVertex AI Recommendation SystemsBigQueryHubSpot
Jan 2024
Dec 2024
Customer Advisory Board Member
HubSpot
Influenced the 12-month time-to-market for the Standard Sandbox via strategic guidance on production metadata, Marketing Hub roadmaps, and CRM UX. Presented growth insights to 350+ global executives at the HubSpot Global Leadership Summit.
Sept 2017
May 2020
Product Manager, Marketplace Integrity & Computer Vision
The MLS (Real Estate Technology)
Beverly Hills, CA · 50K+ daily image uploads · 100K+ professionals
Built CheckMate — a Computer Vision API cutting manual moderation overhead by 80% across 50K+ daily uploads. Built Esta, an NLP conversational AI agent that slashed support ticket volume by 60%. Architected sub-second polygon SQL Spatial search across 15M+ listings, capturing 40% of total platform search volume.
Computer VisionNLP Agents SQL Spatial15M+ Listings
★ HousingWire Real Estate Tech100
Mar 2016
Mar 2017
Founder
VCU.TV
Los Angeles, CA · Web Summit Alpha 2015, Dublin
Founded VCU.TV — a smart device for personalized video discovery built on Android with custom hardware for next-gen codecs. Supported 4K UHD, HD, and 3D. Delivered content via audio-visual analytics tuned to individual viewing preferences, doubling as a smart home connectivity hub and security camera control point.
0-to-1 Hardware/SoftwareAndroid 4K UHDA/V Analytics
★ Web Summit Alpha 2015, Dublin
Mar 2011
Oct 2013
Co-Founder
Exxat, LLC
La Crescenta, CA · B2B SaaS for clinical education management
Bootstrapped a zero-to-one B2B SaaS platform for clinical education. Led go-to-market at national trade shows, secured $55K in early pipeline, and landed the 2 largest US clinical programs as flagship customers — product-market fit from a standing start.
0-to-1 SaaSGo-to-MarketEnterprise Sales
Oct 2010
Sept 2017
Product Manager, SaaS Research Platforms
USC Norris Cancer Research Center
Los Angeles, CA
Scaled platform adoption 17x — from 20 to 350 concurrent clinical trials — by streamlining UX and automating workflows. Structured raw sensor and EDC datasets to train early ML models for detecting neuro-psychological side effects.
Early ML17x GrowthClinical Research
Outside work

I grow underwater ecosystems

Aquascaping is landscape design executed entirely underwater. You're balancing biology, chemistry, fluid dynamics, and aesthetics simultaneously — with zero ability to pause or undo. It genuinely teaches systems thinking in a way most hobbies don't.

Every planted tank is a living product. You define the initial conditions, set constraints, and watch emergence happen. The ecosystem tells you when something's wrong before you've noticed it yourself. That feedback loop maps onto platform work more directly than it has any right to.

I run high-tech CO2-injected planted tanks with a particular obsession with Iwagumi layouts. One rock out of place ruins the entire composition — a useful reminder that precision and restraint matter as much as ambition.

Get in touch

Let's build
something real

Email aavesh.naykodi@gmail.com LinkedIn linkedin.com/in/aaveshnaykodi Substack aavesh.substack.com