Skip to content
View romanmagdalenko's full-sized avatar
🏊‍♂️
I'm vibe swimmer
🏊‍♂️
I'm vibe swimmer

Block or report romanmagdalenko

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
romanmagdalenko/README.md

Roman Magdalenko

Founding Deployment Strategist. I build the pilot-to-production systems for Revenue AI startups, which is the stretch where most of them stall.

26 years in enterprise sales: VP Sales EMEA at ABBYY LS, CCO at RedLab, now founder and CEO at SalesAI. 54 enterprise deployments, and in the last three years 200+ AI pilots launched or rescued.

Based in Valencia, Spain. Open to roles here or remote.

Most of my production code lives in private repos. What I can open, I put here.

Now building

SalesAI - Revenue Intelligence platform. The approach we sell is Managed Revenue Growth: a proprietary LLM sits on a sales knowledge graph of 70+ entity nodes, pulls 19 entities out of every customer conversation, writes them back into the CRM, and turns that into a forecast, a risk list and a coaching plan for each rep. Speech analytics answers how the call went. The graph answers where the money is.

tourdecastles.com - Tour de Castles a tourism demand-distribution platform for Europe's overlooked destinations. It turns castles, cycling routes and local heritage into reasons to travel beyond the usual hotspots, then measures whether that attention becomes actual visits and local economic activity.

The product combines 900+ castle pages, route data, a 23,700-note research vault, Wikidata and a grounded knowledge graph with a behavioural layer that tracks discovery, trip intent and destination engagement.

The long-term metric is not pageviews. It is incremental visitor spending redirected to underserved territories - proving that digital discovery can move tourism demand, not just describe places.

smartsales.ai - my English writing on revenue architecture, and the EN/ES content engine that ships it.

What I actually do

  • Knowledge graphs over messy human data: entity extraction, ontology design, grounding gates, health metrics you can put in CI.
  • Revenue systems: CRM as an event pipeline (firehose into D1), anonymous-to-contact stitching, an AI SDR that runs outbound without a human in the loop.
  • Claude Code as an operating system: skills, memory architecture, crons. One founder running sales, marketing and product on it.
  • Sales methodology as code. MEDDIC, SPICED and BANT are not slides for me, they are schemas.

How the three repositories fit together

They are not a portfolio. They are one argument in three parts: what is true (the knowledge graph), what happened (the behaviour layer), and how one person executes against both (the skills).

All three were built before there was a business model to justify them, which is the point. The reasoning: Designing for Optionality.

Writing

  • smartsales.ai - Revenue Architecture, in English.
  • LinkedIn - shorter notes from the room, not from the deck.

Stack

TypeScript, React, Node. Cloudflare Workers, D1, KV, R2. Python. Postgres. Claude API and MCP servers. Wikidata and SPARQL. GitLab CI, Kubernetes, Supabase.

Reach me

LinkedIn · smartsales.ai · Book 30 minutes

Pinned Loading

  1. claude-gtm-skills claude-gtm-skills Public

    How one founder runs a B2B GTM function on Claude Code: 14 skills, 4 MCP servers, 14 scheduled agents. The architecture, the memory layer and the cron layer, sanitized.

    1

  2. castle-graph-engine castle-graph-engine Public

    A typed knowledge graph over 23,738 research notes, with deterministic publication gates: no auto-merge, every claim sourced, the build compares itself byte for byte.

    1

  3. crm-on-cloudflare crm-on-cloudflare Public

    A first-party CRM on Cloudflare's free tier: Workers, D1, Analytics Engine. Two loads and two stores, identity stitched backwards, consent that survives erasure, and the read-amplification bill tha…

    1

  4. fde-deployment-economics fde-deployment-economics Public

    What an enterprise AI deployment actually costs. A runnable cost model that answers one question before you sign: at what probability of success does this purchase break even? Plus eleven questions…

    Python 1