Yash Verma

• I was the top on-call engineer in the company. I resolved tickets in the shortest time. I resolved the most tickets.
• Weekly active teachers increased by more than 10%. Registrations increased by 5%. Server-side rendering and performance work caused these increases.
• I built the question types, the lesson editor, and the gameplay. The platform still gets revenue from these features.
• I added canary deployments, application performance monitoring, real user monitoring, and analytics. The system became more stable.

Search on Postgres

I moved the OpenSearch indexes to Postgres. Search uses full text search and pagination. Kafka and Debezium CDC keep the data current.

• This change saved $270k.

Postgres · Kafka · Debezium · OpenSearch

Less data transfer

I reduced data transfer with zstd and gzip compression. I also added role-based projections.

• This change saved $100k.

zstd · gzip

BigQuery streaming

I enabled the BigQuery streaming API. Better batch sizes and compression reduced the number of pods and the data transfer.

• This change saved $120k.

BigQuery

Data analysis platform

I built the Waygent analytics platform. It has a session explorer, standard A/B tests, and release-health checks.

• Registration funnels increased by 2%.
• The game creation rate increased by 0.5%.
• A/B test release time decreased from 2 weeks to 1 week.

Next.js · React · Golang · BigQuery · Kinesis · Firehose

Analysis chat agent

I built a chat agent that answers analysis questions in normal language. The questions include daily active users, A/B test results, click rates, and impressions.

Python · Agno · BigQuery · ChromaDB

Analytics pipeline

I built an analytics pipeline with the beacon API, Golang, and Kinesis. The pipeline accepts a high request rate.

• The peak rate was 300k requests each minute.
• The service level was 99.99%.

Golang · Kinesis · Firehose · SQS · BigQuery

Analytics client

I built an analytics SDK for React, Next, Nuxt, and Vue. The SDK records sessions and referrers. It also adds events from the design system.

• Data loss decreased by 15%.
• Automatic events made important funnels easier to track.
• All microfrontends adopted the SDK.

TypeScript

Server-side rendering

I added server-side rendering with Nuxt and a design system. Organic traffic increased after this work.

• User registration from organic traffic increased by 15%.

Vue · Nuxt · Tailwind

Static page pipeline

I built a pipeline for static pages. The pipeline serves HTML from S3. Static pages do not call the server.

• Time to first byte improved by 12%.
• Largest contentful paint improved by 31%.

Golang · S3 · SQS · Nuxt

Canary deployments

I deployed the frontend and the backend with Jenkins and AWS CodeDeploy. Deployments became more reliable.

• P0 incidents decreased by 83%.

CodeDeploy · ECS · Lambda

ECS auto-scaling

I added an AWS capacity provider and ECS auto-scaling. This change decreased the AWS cost.

• AWS cost decreased by 10%.

ECS

Website performance

I decreased the page load time. I used code splitting, vendor chunks, and stable chunk hashes.

• Page load time at p75 decreased by 63%.
• New user activation increased by 1%.

Vue · Vite

Design system

I built a design system with Tailwind CSS. I added the design system to the product.

Build time

I decreased the build time. I used a Docker image cache in ECR.

• Build time decreased by 15%.

ECR · Docker

Tool migration

I moved the product to Turborepo, Vite, and Pinia. The move made builds faster for developers.

• Build time decreased by 50%.

Turborepo · Vite · Docker · Pinia · Jenkins

Content editor

I built rich-text features for the content editor.

Quiz and lesson editor

I built a new quiz editor and a new lesson editor with Pub/Sub and Zod. The editors became faster and more reliable.

• Bug reports decreased by 98%.

AI features

I added AI features. One feature generates slides. Lesson use increased after these features.

• Lesson adoption increased by 2.3%.

Data ingestion pipeline

I built a data pipeline with Temporal. The pipeline ingests unique standards data.

• The pipeline processed more than 500,000 rows.

Temporal

Google Drive import

I added Google Drive import. The AI features use the imported media.

• 56% of platform content came from this integration.

Subscription control

I added access control on the client for subscriptions.

Search results

I made search faster. I also changed the search interface. The search success rate increased.

• Search response at p99 improved by 23%.
• The search success rate increased by 1.2%.

Progressive web app

I built a progressive web app with React, TypeScript, GraphQL, and Redux.

Metadata components

I built components from GraphQL metadata. These components decreased the time to production.

Analytics APIs

I built analytics APIs with Cube.js, PostgreSQL, and Node.js.

Website performance

I made the website faster. I used a cache, pagination, and fewer network calls.

• Website performance increased by 60%.
• The Lighthouse score increased.

Progressive web app

I built a progressive web app with React, Node, and MySQL.

Cloud infrastructure

I operated AWS EC2, RDS, and S3. These services hosted the app, the database, and the files.

Offline use

I added offline use with IndexedDB. I also added live patient diagnosis with AWS services.

• The Lighthouse score increased by 40%.

Chat components

I built reusable components for the chat feature of a job portal.

Unit tests

I added unit tests with Cypress.

Analytics dashboard

I built an analytics dashboard with React. The dashboard shows user API performance.

API connection

I connected APIs that track performance.

Payments

I added payments with Stripe.