Olga Ostrovskiy
By day I am a Tech Lead at AstraZeneca, building AI platforms for drug discovery. By night I build My Storyland: an app that turns the book you just finished into a trip you can walk.
I like helping people put AI to work, whether that is an engineer shipping agents or a friend who uses ChatGPT as a search engine. How to bring it in, and how to check it did what you asked. Here: what I build, what I measured, and the occasional thing that did not work. No schedule.
Just back from Morocco.
Start here
First time? These three.
- Try My StorylandGive it a book you love and see where it sends you. Everything else here is downstream of this.
- The DevFest DC talkThirty minutes on how the agents turn a book into a trip, and how I check the trip is any good. Slides included.
- The grocery list that fills itselfThe smallest useful thing I have built with AI, and the easiest to copy.
Projects
My Storyland live
Give it a book, get a trip. A multi-agent system reads the novel, finds the places behind it, and builds a walkable itinerary with the book-to-place connection explained at every stop.
Everyday skills active
Small automations for a family, built with Claude and connected tools. A grocery list that fills itself from what anyone says during the week; a wardrobe that picks the outfit.
Sport Python learning active
Python notebooks for kids who love sports, on open NFL data, in Google Colab. Nothing to install.
ScoreZone done
A live sports scores app (NBA, MLB, golf) with standings, injuries and news. Built in a one-hour session with my son, from his idea.
Notes
- Python notebooks for a kid who loves football
- An evening of AI, on vacation
- Shaping the build
- I tried the course's code graph on my own backend
- DevFest DC, my first time as a speaker
Ask me about
- Which AI to try for which task, and how to tell whether it did the job.
- Writing a prompt on a walk: one or two miles talking to the model like a friend, then asking it to turn the conversation into prompts.
- Agents with access to tools, not sandboxes, and the guardrails that keep them honest.
- Checking a model's answer: evaluation sets, a judge calibrated against people, and a cost you can defend.