At Geomir, I independently developed a microservice connecting autonomous drone ports to the AI-driven agritech platform. I built its PostgreSQL data model and integration logic to automate mission scheduling and send aerial imagery to AWS S3 for ML-powered crop forecasting; the solution is undergoing pilot testing.
I migrated farm machinery GPS telemetry ingestion to real-time streaming across partner protocols, reducing latency from several hours to minutes. I also added trajectory-based filtering that removed about 10% of outliers and stale points.
At Geomir, I fixed batch synchronization and execution limits in an ML route-optimization pipeline, restoring reliable route processing. I also reduced long-running API response times from 30–60 seconds to 5–10 seconds by optimizing SQL queries and parallelizing S3 downloads.
For Breaking Barriers, I developed the backend API and scheduled job queues for a Telegram bot that sends daily tasks to children. I deployed the service on Yandex Cloud using Docker and documented its API with Swagger / OpenAPI.
