Ulad Ramanovich
@uramanovich.bsky.social
Senior Fullstack Engineer passionate about TypeScript/JavaScript ecosystems.
Sharing pragmatic tech tips & exploring AI-powered development workflows.
Let's build better software together! 🚀
Sharing pragmatic tech tips & exploring AI-powered development workflows.
Let's build better software together! 🚀
The `nyc` npm package lets you merge coverage reports from different test jobs into one comprehensive report. Here's how it works:
April 26, 2025 at 9:58 AM
The `nyc` npm package lets you merge coverage reports from different test jobs into one comprehensive report. Here's how it works:
Tired of slow Jest tests? I cut my test suite time by 67% with these two simple tricks:
1. Add `maxWorkers=50%` to utilize available CPU
2. Implement `shard` to split and run tests in parallel across multiple jobs
The big challenge? Collecting coverage data from separate jobs. Solution below
1. Add `maxWorkers=50%` to utilize available CPU
2. Implement `shard` to split and run tests in parallel across multiple jobs
The big challenge? Collecting coverage data from separate jobs. Solution below
April 26, 2025 at 9:58 AM
Tired of slow Jest tests? I cut my test suite time by 67% with these two simple tricks:
1. Add `maxWorkers=50%` to utilize available CPU
2. Implement `shard` to split and run tests in parallel across multiple jobs
The big challenge? Collecting coverage data from separate jobs. Solution below
1. Add `maxWorkers=50%` to utilize available CPU
2. Implement `shard` to split and run tests in parallel across multiple jobs
The big challenge? Collecting coverage data from separate jobs. Solution below
What is your next action when you have an incident and the last PR in the list this one?
February 19, 2025 at 6:30 AM
What is your next action when you have an incident and the last PR in the list this one?
Confusing by word 'token' when working with AI? Me too 🤔
I spent time understanding how it's calculated, here's simple math:
- Token ≠ word
- 1 word ≈ 1.3token
- 1 tweet ≈ 35 tokens
- Emojis cost extra tokens
shorter words = fewer tokens = lower costs
I spent time understanding how it's calculated, here's simple math:
- Token ≠ word
- 1 word ≈ 1.3token
- 1 tweet ≈ 35 tokens
- Emojis cost extra tokens
shorter words = fewer tokens = lower costs
February 14, 2025 at 2:40 PM
Confusing by word 'token' when working with AI? Me too 🤔
I spent time understanding how it's calculated, here's simple math:
- Token ≠ word
- 1 word ≈ 1.3token
- 1 tweet ≈ 35 tokens
- Emojis cost extra tokens
shorter words = fewer tokens = lower costs
I spent time understanding how it's calculated, here's simple math:
- Token ≠ word
- 1 word ≈ 1.3token
- 1 tweet ≈ 35 tokens
- Emojis cost extra tokens
shorter words = fewer tokens = lower costs
4/5 Testing becomes trivial:
- Unit tests: mock DataService
- E2E: use TestDatabaseService with Fishery
- No need to mock individual repositories
- Clean, reusable test data
- Unit tests: mock DataService
- E2E: use TestDatabaseService with Fishery
- No need to mock individual repositories
- Clean, reusable test data
February 10, 2025 at 3:17 PM
4/5 Testing becomes trivial:
- Unit tests: mock DataService
- E2E: use TestDatabaseService with Fishery
- No need to mock individual repositories
- Clean, reusable test data
- Unit tests: mock DataService
- E2E: use TestDatabaseService with Fishery
- No need to mock individual repositories
- Clean, reusable test data
3/5 Core components:
1. Base Repository - shared database operations
2. Entities - database structure
3. DataService - single entry point
4. Factories - test data generation
1. Base Repository - shared database operations
2. Entities - database structure
3. DataService - single entry point
4. Factories - test data generation
February 10, 2025 at 3:17 PM
3/5 Core components:
1. Base Repository - shared database operations
2. Entities - database structure
3. DataService - single entry point
4. Factories - test data generation
1. Base Repository - shared database operations
2. Entities - database structure
3. DataService - single entry point
4. Factories - test data generation
2/5 The key is centralizing all database logic in one place.
Create a data-access module that:
• Acts as a single source of truth
• Handles all database operations
• Provides clean testing interfaces
• Manages multiple databases
Create a data-access module that:
• Acts as a single source of truth
• Handles all database operations
• Provides clean testing interfaces
• Manages multiple databases
February 10, 2025 at 3:17 PM
2/5 The key is centralizing all database logic in one place.
Create a data-access module that:
• Acts as a single source of truth
• Handles all database operations
• Provides clean testing interfaces
• Manages multiple databases
Create a data-access module that:
• Acts as a single source of truth
• Handles all database operations
• Provides clean testing interfaces
• Manages multiple databases
1/5 How to manage database access in monorepo with NestJS?
I faced this problem a few times:
• Duplicated database logic
• Multiple connection configurations
• Inconsistent testing patterns
• No clear structure
Let me explain how I fixed this with the Data-Access Pattern 🧵
#NestJS #nodejs #webdevb
I faced this problem a few times:
• Duplicated database logic
• Multiple connection configurations
• Inconsistent testing patterns
• No clear structure
Let me explain how I fixed this with the Data-Access Pattern 🧵
#NestJS #nodejs #webdevb
February 10, 2025 at 3:17 PM
This is env management I bring for each project. Don't rely on process.env directly - use Zod validation instead.
Type-safe + clear errors = no more missing configs
#TypeScript #WebDev #nodejs
Type-safe + clear errors = no more missing configs
#TypeScript #WebDev #nodejs
January 29, 2025 at 4:26 PM
This is env management I bring for each project. Don't rely on process.env directly - use Zod validation instead.
Type-safe + clear errors = no more missing configs
#TypeScript #WebDev #nodejs
Type-safe + clear errors = no more missing configs
#TypeScript #WebDev #nodejs
The 70/30 rule of AI coding: While AI can scaffold 70% of your code, the crucial 30% requires deep engineering expertise.
Best practice: Use AI as a drafting tool, maintain atomic conversations, and own your code through meaningful review.
#AI #SoftwareDevelopment
Best practice: Use AI as a drafting tool, maintain atomic conversations, and own your code through meaningful review.
#AI #SoftwareDevelopment
January 24, 2025 at 5:33 PM
The 70/30 rule of AI coding: While AI can scaffold 70% of your code, the crucial 30% requires deep engineering expertise.
Best practice: Use AI as a drafting tool, maintain atomic conversations, and own your code through meaningful review.
#AI #SoftwareDevelopment
Best practice: Use AI as a drafting tool, maintain atomic conversations, and own your code through meaningful review.
#AI #SoftwareDevelopment
Multiple tabs sending same requests? Been there 🔧
Solution: Leader Election pattern!
- Tabs form a team
- Pick ONE leader tab
- Only leader makes API calls
- If leader closes, new one steps up
Used broadcast-channel lib. Now instead of 5 requests → just 1
#frontend #javascript
Solution: Leader Election pattern!
- Tabs form a team
- Pick ONE leader tab
- Only leader makes API calls
- If leader closes, new one steps up
Used broadcast-channel lib. Now instead of 5 requests → just 1
#frontend #javascript
January 20, 2025 at 7:50 PM
Multiple tabs sending same requests? Been there 🔧
Solution: Leader Election pattern!
- Tabs form a team
- Pick ONE leader tab
- Only leader makes API calls
- If leader closes, new one steps up
Used broadcast-channel lib. Now instead of 5 requests → just 1
#frontend #javascript
Solution: Leader Election pattern!
- Tabs form a team
- Pick ONE leader tab
- Only leader makes API calls
- If leader closes, new one steps up
Used broadcast-channel lib. Now instead of 5 requests → just 1
#frontend #javascript
Multiple tabs sending same requests? Been there 🔧
Solution: Leader Election pattern!
- Tabs form a team
- Pick ONE leader tab
- Only leader makes API calls
- If leader closes, new one steps up
Used broadcast-channel lib. Now instead of 5 requests → just 1
#frontend #javascript
Solution: Leader Election pattern!
- Tabs form a team
- Pick ONE leader tab
- Only leader makes API calls
- If leader closes, new one steps up
Used broadcast-channel lib. Now instead of 5 requests → just 1
#frontend #javascript
January 20, 2025 at 4:00 PM
Multiple tabs sending same requests? Been there 🔧
Solution: Leader Election pattern!
- Tabs form a team
- Pick ONE leader tab
- Only leader makes API calls
- If leader closes, new one steps up
Used broadcast-channel lib. Now instead of 5 requests → just 1
#frontend #javascript
Solution: Leader Election pattern!
- Tabs form a team
- Pick ONE leader tab
- Only leader makes API calls
- If leader closes, new one steps up
Used broadcast-channel lib. Now instead of 5 requests → just 1
#frontend #javascript
Discovered today Docker Compose profiles and was surprised.
Instead of juggling multiple docker-compose files for prod/dev/test, just add "profiles: [dev]" to your services and run:
docker compose --profile dev up
Perfect for managing different environments in a single file!
#docker
Instead of juggling multiple docker-compose files for prod/dev/test, just add "profiles: [dev]" to your services and run:
docker compose --profile dev up
Perfect for managing different environments in a single file!
#docker
January 7, 2025 at 6:16 PM
Discovered today Docker Compose profiles and was surprised.
Instead of juggling multiple docker-compose files for prod/dev/test, just add "profiles: [dev]" to your services and run:
docker compose --profile dev up
Perfect for managing different environments in a single file!
#docker
Instead of juggling multiple docker-compose files for prod/dev/test, just add "profiles: [dev]" to your services and run:
docker compose --profile dev up
Perfect for managing different environments in a single file!
#docker
Tired of TypeScript being "stupid" with filter()?
When you filter out undefined, does TS still think they might be there?
🪄 Magic solution: Add "is" type check!
#typescript #webdev
When you filter out undefined, does TS still think they might be there?
🪄 Magic solution: Add "is" type check!
#typescript #webdev
January 7, 2025 at 7:47 AM
Tired of TypeScript being "stupid" with filter()?
When you filter out undefined, does TS still think they might be there?
🪄 Magic solution: Add "is" type check!
#typescript #webdev
When you filter out undefined, does TS still think they might be there?
🪄 Magic solution: Add "is" type check!
#typescript #webdev
I prefer Claude over ChatGPT for both its superior code assistance and projects feature (Claude Pro).
With projects, you can create custom prompts with specific output requirements - perfect for recurring tasks like architecture planning or article title generation
With projects, you can create custom prompts with specific output requirements - perfect for recurring tasks like architecture planning or article title generation
January 1, 2025 at 2:42 PM
I prefer Claude over ChatGPT for both its superior code assistance and projects feature (Claude Pro).
With projects, you can create custom prompts with specific output requirements - perfect for recurring tasks like architecture planning or article title generation
With projects, you can create custom prompts with specific output requirements - perfect for recurring tasks like architecture planning or article title generation
🔄 Solved my NestJS DB headaches by creating a Data Access module - a dedicated layer that handles all database operations, migrations, and models. It manages both Postgres & MongoDB, centralizes connections, and makes testing easy.
Writing an article to share my complete setup! 📝
Writing an article to share my complete setup! 📝
December 19, 2024 at 7:55 AM
🔄 Solved my NestJS DB headaches by creating a Data Access module - a dedicated layer that handles all database operations, migrations, and models. It manages both Postgres & MongoDB, centralizes connections, and makes testing easy.
Writing an article to share my complete setup! 📝
Writing an article to share my complete setup! 📝
It depends on your application. This is what I added:
- I send to server "devMode" flag and remove any restrictions (player count, for starting the game, etc.)
- Debug information, like "row state". It's helpful when you're testing release build
Here is an example of my debug panel:
- I send to server "devMode" flag and remove any restrictions (player count, for starting the game, etc.)
- Debug information, like "row state". It's helpful when you're testing release build
Here is an example of my debug panel:
December 8, 2024 at 2:29 PM
It depends on your application. This is what I added:
- I send to server "devMode" flag and remove any restrictions (player count, for starting the game, etc.)
- Debug information, like "row state". It's helpful when you're testing release build
Here is an example of my debug panel:
- I send to server "devMode" flag and remove any restrictions (player count, for starting the game, etc.)
- Debug information, like "row state". It's helpful when you're testing release build
Here is an example of my debug panel:
1. Choose not so important screen. I chosen language selector view
2. Add hidden button on the bottom
3. Create logic that activate `isDevMode` flag when taped 10 times
4. Show indicator that dev mode enabled
Voilà ✨
2. Add hidden button on the bottom
3. Create logic that activate `isDevMode` flag when taped 10 times
4. Show indicator that dev mode enabled
Voilà ✨
December 8, 2024 at 11:05 AM
1. Choose not so important screen. I chosen language selector view
2. Add hidden button on the bottom
3. Create logic that activate `isDevMode` flag when taped 10 times
4. Show indicator that dev mode enabled
Voilà ✨
2. Add hidden button on the bottom
3. Create logic that activate `isDevMode` flag when taped 10 times
4. Show indicator that dev mode enabled
Voilà ✨
Building mobile app? Pro tip: Add a hidden dev mode for App Store reviews and testing
Implement a secret tap combo in a corner, show a dev mode indicator, and add debug controls. Makes testing simple for reviewers, plus helps verify features work 🔍
Here is how I did this 🧵
#buildinpublic #swiftui
Implement a secret tap combo in a corner, show a dev mode indicator, and add debug controls. Makes testing simple for reviewers, plus helps verify features work 🔍
Here is how I did this 🧵
#buildinpublic #swiftui
December 8, 2024 at 11:05 AM
Building mobile app? Pro tip: Add a hidden dev mode for App Store reviews and testing
Implement a secret tap combo in a corner, show a dev mode indicator, and add debug controls. Makes testing simple for reviewers, plus helps verify features work 🔍
Here is how I did this 🧵
#buildinpublic #swiftui
Implement a secret tap combo in a corner, show a dev mode indicator, and add debug controls. Makes testing simple for reviewers, plus helps verify features work 🔍
Here is how I did this 🧵
#buildinpublic #swiftui
🎮 Wrapped up 7 intense days building "Spot the Stranger" - party game where players find outsider who doesn't know the secret profession!
Completed core features: lobbies, gameplay, scoring & real-time communication. All while working full-time!
Next stop: Bug fixing and App Store 🚀
#buildinpublic
Completed core features: lobbies, gameplay, scoring & real-time communication. All while working full-time!
Next stop: Bug fixing and App Store 🚀
#buildinpublic
December 3, 2024 at 6:36 PM
🎮 Wrapped up 7 intense days building "Spot the Stranger" - party game where players find outsider who doesn't know the secret profession!
Completed core features: lobbies, gameplay, scoring & real-time communication. All while working full-time!
Next stop: Bug fixing and App Store 🚀
#buildinpublic
Completed core features: lobbies, gameplay, scoring & real-time communication. All while working full-time!
Next stop: Bug fixing and App Store 🚀
#buildinpublic
🎮 Spot the Stranger - Day 6/7
Added new mechanics after testing:
The Stranger can now win by guessing the profession. Making Citizens think twice about their questions! Also, a new voting system is in place to catch the Stranger.
#buildinpublic
Added new mechanics after testing:
The Stranger can now win by guessing the profession. Making Citizens think twice about their questions! Also, a new voting system is in place to catch the Stranger.
#buildinpublic
November 30, 2024 at 11:39 AM
🎮 Spot the Stranger - Day 6/7
Added new mechanics after testing:
The Stranger can now win by guessing the profession. Making Citizens think twice about their questions! Also, a new voting system is in place to catch the Stranger.
#buildinpublic
Added new mechanics after testing:
The Stranger can now win by guessing the profession. Making Citizens think twice about their questions! Also, a new voting system is in place to catch the Stranger.
#buildinpublic
Just discovered raycast app for mac - perfect tool for window management and running commands 👀
Beautiful UI that lets you run commands without memorizing shortcuts.
Replaced my skhd + yabai setup, and it's as simple as installing from brew.
Beautiful UI that lets you run commands without memorizing shortcuts.
Replaced my skhd + yabai setup, and it's as simple as installing from brew.
November 28, 2024 at 9:35 PM
Just discovered raycast app for mac - perfect tool for window management and running commands 👀
Beautiful UI that lets you run commands without memorizing shortcuts.
Replaced my skhd + yabai setup, and it's as simple as installing from brew.
Beautiful UI that lets you run commands without memorizing shortcuts.
Replaced my skhd + yabai setup, and it's as simple as installing from brew.