Co-founder, Eloqo · Barcelona, Spain

I build Eloqo, the AI speech coach in your pocket.

I'm Ilia Gordeev, a senior Kotlin engineer and co-founder of Eloqo. I build the product end to end: the Android and iOS apps, the Firebase backend and the Gemini review pipeline. Before Eloqo, I spent six years on the backend of an oncology patient-monitoring platform.

What I'm building

Eloqo is a gym for your speaking muscles.

Eloqo is a mobile AI speech coach. You record a short speaking exercise, such as an articulation drill, a monologue, shadowing or reading aloud. It scores clarity, fluency, vocabulary and confidence against the transcript, then names one thing to fix next.

  1. 01

    Record a short take. One to three minutes, from tongue twisters to retelling a text from memory.

  2. 02

    See instant numbers. Pace, filler words and hesitation pauses are calculated on the device, free and without an account.

  3. 03

    Ask for a coach review. Gemini on Vertex AI scores the take, annotates the transcript and suggests the next exercise.

Eloqo home screen with average score and today's plan Eloqo recording screen during a monologue Eloqo session review with score breakdown
Android + iOS
One Kotlin Multiplatform codebase
In beta testing
Open on TestFlight, closed on Google Play
English + Russian
Practice languages today
Two co-founders
Engineering and speech science

Built on Google Cloud from day one.

Every part of Eloqo outside the phone runs on Google Cloud and Firebase. The whole backend is serverless, so two people can run it.

Vertex AI Gemini API
Coach reviews, through Firebase AI Logic
Cloud Firestore
Sessions, scores and cross-device sync
Cloud Storage
Voice recordings
Cloud Functions
Subscription checks, AI quotas, data cleanup
BigQuery
Product metrics exported from Firestore
Firebase Auth and App Check
Sign-in and abuse protection
Firebase Hosting
eloqo.org and the admin console
Remote Config and Crashlytics
Quota tuning and stability monitoring
The team

I co-founded Eloqo with Julia Lysova, a PhD candidate in computational linguistics and cognitive science. I lead engineering; Julia leads the speech methodology.

Meet the team on eloqo.org

Experience

Systems that cannot fail, then a product of my own.

Mar 2026 – Present

Co-founder, Eloqo

Product and engineering for a mobile AI speech coach: Kotlin Multiplatform apps for Android and iOS, a serverless Firebase backend, the Gemini review pipeline, subscriptions on Google Play and the App Store, and the website and admin console.

6 years

Senior backend engineer, Oncor

Backend of an oncology patient-monitoring platform with heavy integrations with national and regional health systems.

  • Built medical document exchange on HL7 CDA, over REST, SOAP, Protobuf and XML/XSD, on PostgreSQL and OrientDB.
  • Led the Java to Kotlin migration, which cut production bugs by about 20%.
  • Built an LLM microservice for government contract analysis: batch inference, message-driven processing, containerized deployment.

Stack

Kotlin, front to back.

Mobile
Kotlin Multiplatform, Compose Multiplatform, Room, Koin
Cloud
Firebase, Cloud Functions (TypeScript), Firestore, BigQuery, Vertex AI
Backend
Kotlin, Java, Spring Boot, Ktor, Exposed, Kafka
Integration
REST, SOAP, Protobuf, XML/XSD, HL7 CDA
Data
PostgreSQL, OrientDB, SQL
AI
Gemini, Python, vLLM, local inference
Systems and delivery
Rust, GTK 4, Linux, Docker, GitHub Actions, Gradle

Also building

Open source.

Addict, a pop-up dictionary for Linux

Select a word, press a hotkey, and see its meanings, synonyms and pronunciation, with the sense that fits the sentence. Rust and GTK 4, offline dictionaries, OCR for text in images, optional DeepL. Ships as Flatpak and AppImage.

Let's talk about Eloqo.

For anything about the product, partnerships or the engineering behind it, write to me directly.