Tech Lead, Backend and Data Platform
Umair Mirza
I lead a team of 10+ engineers and build geospatial data platforms on AWS: Airflow pipelines for satellite imagery, a TB-scale geospatial data lake with Apache Iceberg and Trino, and Go services.
- Lahore, Pakistan · Farmdar (Series A agritech)
- Open to remote roles and relocation (Netherlands, Ireland, Germany, Luxembourg)
- engineers led
- 10+engineers led
- API requests a day
- 100K+API requests a day
- geospatial data on S3
- TB-scalegeospatial data on S3
- YieldPro revenue
- $200K+YieldPro revenue
- of acres covered
- Millionsof acres covered
- lower AWS spend
- 62%lower AWS spend
About
What I do
I'm the tech lead for backend and data platform at Farmdar, a Series A agritech company. I lead a team of 10+ engineers and own the platform end to end, from the Airflow pipelines that process satellite imagery to the TB-scale data lake and the Go backend that serves 100K+ requests a day.
Our products cover millions of acres, and the API is used in production by FFC, Corteva, Engro and Kubota. I enjoy data platforms, geospatial problems and helping a team ship good systems.
I'm open to remote roles and to relocating to the Netherlands, Ireland, Germany or Luxembourg.
Experience
Where I've worked
Farmdar
Series A agritech · Lahore, Pakistan
Tech Lead, Backend and Data Platform
Feb 2025 - Present- Lead a team of 10+ engineers; own architecture, code review and delivery across backend, data and infrastructure.
- Designed the lakehouse layer of a TB-scale geospatial data lake: Apache Iceberg tables on AWS S3 Tables and Trino as one SQL layer over the lake and the PostGIS production database.
- Built OAuth authentication on AWS Cognito, used by every Farmdar application, and consolidated separate services into one Go backend handling 100K+ daily requests, with P95 latency down 40%.
- Automated the 3 m Planet imagery pipeline (plant health, stress and productivity-zone maps), cutting delivery from weeks to hours; built 10+ Airflow DAGs and a custom Airflow-Kubernetes operator, cutting manual work by 70%.
- Own infrastructure as code with Terraform, Helm and GitHub Actions, cutting deployment time by 60%; led a cost review that cut AWS spend by 62%. Launched a production API adopted by FFC, Corteva, Engro and Kubota, and built a crop water requirement tool for the FAO.
Software Engineer II
Jan 2024 - Jan 2025- Led development of YieldPro, a crop-monitoring platform built from scratch, generating $200K+ in revenue and serving 50K+ farmers.
- Automated satellite image analysis with Apache Airflow, improving data throughput 3x.
- Introduced role-based access control (RBAC) and built a PostGIS tile server for vector and raster tiles.
- Delivered an Angular geospatial app for real-time map analytics, cutting user task completion time by 50%.
Software Engineer
Mar 2023 - Dec 2023- Designed MongoDB schemas for geospatial and agronomic datasets and built an internal React console for managing map layers.
- Refactored a legacy React/Redux and Kepler codebase and supported its migration to Mapbox GL JS.
Tectiers GmbH
Fintech, part-time remote contract · Vienna, Austria
Software Engineer
Oct 2022 - Sep 2023- Built Apache Airflow pipelines for BMW's enterprise reporting platform, automating data extraction and weekly database restores across environments.
- Built a fintech analytics platform with Angular, Django and PostgreSQL.
- Wrote a Python transpiler that converts Excel/VBA logic into native Excel functions with 95% functional parity.
FAST-NUCES
University · Lahore, Pakistan
Teaching Assistant
Feb 2022 - May 2023- Mentored 100+ students in Android (Kotlin) and distributed computing (MPI, OpenMP).
Selected work
What I've built
One SQL layer, TB-scale data
Geospatial data lake and query layer
An Apache Iceberg lakehouse on AWS S3 Tables with Trino as a single SQL layer over the lake and the PostGIS production database, signed in through Cognito.
Analysts and ML engineers query lake tables and live production data from one place, with access controlled by Cognito groups.
Read the case study
Weeks to hours
Satellite imagery pipelines
Production Airflow on Kubernetes for satellite imagery: Planet 3 m ingestion delivered straight to S3, an operator that runs each Jupyter notebook in its own pod, and the reporting pipelines on top.
Throughput up 3x, manual work down 70%, Planet imagery delivery from weeks to hours, and AWS spend down 62%.
Read the case study
$200K+ revenue
YieldPro
A crop-monitoring platform built from scratch: field subscriptions in PostGIS, Airflow pipelines that produce per-field crop-health reports, vector tiles served from the database, and an Angular map interface.
$200K+ in revenue and 50K+ farmers served.
Read the case study
4 backends into 1
Unified Go backend and Cognito sign-in
Merged four product backends into one Go service and moved every application to AWS Cognito sign-in, with a scoped partner API at the edge.
100K+ daily requests, P95 latency down 40%, and a production API adopted by FFC, Corteva, Engro and Kubota.
Read the case study
FAO, yield, CropScan
CropScan in progressCrop intelligence: from models to products
The engineering that turns crop models into products: the FAO crop-pattern planning platform, a pixel-level corn and soybean yield pipeline, and CropScan, which I'm building now.
Models built by the data science team run as scheduled pipelines and ship to clients and the FAO as products, not notebooks.
Read the case study
Also
Writing
Notes from production
· 4 min read
Ten million S3 list requests from one Airflow run
A single crop-monitoring run listed an S3 folder about ten million times in one day. The cause was one GDAL default, and the fix was four environment variables set in the right place.
· 7 min read
Moving every Farmdar app to Cognito without a big-bang cutover
How we replaced a home-grown JWT login with AWS Cognito across web and mobile apps in phases: dual-mode tokens behind a feature flag, a hosted UI, and a provisioning bug we only found in testing.
Skills
Stack
- Languages
- Go
- Python (Airflow, FastAPI, Django, GDAL, Rasterio, NumPy, Pandas)
- TypeScript (Angular, React)
- SQL
- Data and geospatial
- Apache Airflow
- Apache Iceberg
- AWS S3 Tables
- Trino
- Spark
- PostgreSQL
- PostGIS
- MongoDB
- Google Earth Engine
- TensorFlow
- Mapbox GL
- Cloud and infrastructure
- AWS (EKS, RDS, S3, Cognito, IAM)
- Kubernetes
- Terraform
- Helm
- GitHub Actions
Education
Background
BS Computer Science, FAST-NUCES, Lahore
2023- Dean's List, four consecutive semesters
- Teaching Assistant
Contact
Get in touch
The fastest way to reach me is email. Remote, UTC+5, overlaps European working hours.