01
Customer products
Turn customer requirements, feedback, and field constraints into dashboards, forecasting, optimizers, and customer-specific rollouts.
Dashboard · Forecasting · Linear programming · React
Shobhit Pandey · Singapore
Senior Software Engineer
Customer-facing products, cloud platforms, and reliable data systems built for real operations.
I build fast analytics products.
I work from customer requirements through architecture, rollout, and production support. My experience spans Python, Java, React, distributed data pipelines, observability, and cloud delivery across AWS, Azure, and GCP.
Based in Singapore · Open to international relocation and travel
Operating path
Live- 01Customer
- 02Product
- 03Platform
- 04Data
- 05Operations
- 06Outcome
7+ years
Software engineering
AWS · Azure · GCP
Production cloud ownership
About 2 TB/week
Data processed
99.9%
Uptime sustained
01 · impact
Software ownership across the whole operating path
My work crosses product, platform, data, and delivery boundaries. I stay close to the customer problem while taking responsibility for how the system behaves in production.
02
Cloud and platform
Design, deploy, operate, and migrate production workloads across AWS, Azure, and GCP.
Docker · Kubernetes · CI/CD · Security
03
Data reliability
Build replay-safe processing systems with idempotency, distributed locks, retries, recovery checkpoints, and operational alerts.
Prefect · Python · Event-driven systems · Observability
04
Delivery leadership
Lead planning, design reviews, code reviews, mentoring, and implementation across product, pipeline, automation, and operations.
5 to 10 engineers and interns mentored over time
02 · selected impact
Decisions that changed production outcomes
The strongest results came from understanding the operating constraint first, then changing the right part of the system.
01Production outcome
Turning field constraints into a practical harvesting plan
Result
A customer-facing workflow that combines dashboard visualization, forecasting, and an optimizer to produce practical harvesting schedules.
Operating flow
Customer feedback and real-world constraints stay connected to the final schedule, rather than disappearing inside a technical model.
Problem
A customer needed more than processed data. They needed a plan they could use in day-to-day field operations.
Constraint
The schedule had to respect real operating constraints identified through customer feedback and working sessions.
Decision
Build the workflow iteratively, then add forecasting and formulate the scheduling problem as a linear-programming optimizer.
System path
Valid schedule
- 01Customer feedback
- 02Dashboard
- 03Forecast
- 04Constraints
- 05Optimized schedule
- Python
- Django
- React
- Data visualization
- Linear programming
02Production outcome
Making a slow analytics product feel immediate
Result
Reduced dashboard page loads to under 3 seconds and API p95 latency to 150-200 milliseconds while delivering multi-gigabyte heatmap layers.
- Page loadsfrom 15-20 seconds
- Under 3s
- API p95 latencyfrom over 60 seconds
- 150-200ms
- Heatmap layers via CDN
- 3-5 GB
Problem
Slow pages and minute-long API calls made an agricultural analytics dashboard difficult to use.
Constraint
The product also needed to deliver 3 to 5 GB heatmap layers without a full rewrite.
Decision
Profile real query patterns, add PostgreSQL indexes, cache high-value endpoints through Nginx, and move large assets to CDN-backed delivery with pagination and asynchronous loading.
System path
Optimized path
- 01React
- 02Django API
- 03Nginx cache
- 04PostgreSQL
- 05CloudFront
- Django
- React
- PostgreSQL
- Nginx
- AWS
- CloudFront
03Production outcome
Turning a fragile pipeline into a recoverable one
Result
Raised on-time delivery to 98%, reduced failures to under 5%, and sustained 99.9% uptime at about 2 TB of processing per week.
- Failure ratefrom about 40%
- Under 5%
- On-time deliveryfrom about 60%
- 98%
- Issue identification
- Under 2 min
- Uptime sustained
- 99.9%
Problem
Pipeline failures created repeated manual recovery work and put customer delivery targets at risk.
Constraint
Long-running work had to recover safely without duplicating processing or restarting every completed stage.
Decision
Make stages replay-safe, key work by capture, client, and site identifiers, and add distributed locks, bounded retries, recovery checkpoints, and operational alerts.
System path
Resume safe
- 01Event
- 02Prefect run
- 03Replay-safe stages
- 04Checkpoint
- 05Observability
- 06Delivery
- Python
- Prefect
- Azure
- Elastic
- Sentry
- Supabase
- Metabase
- n8n
04Production outcome
Moving production workloads while delivery continued
Result
Completed a major Azure-to-GCP migration in about one month while production data delivery continued.
- End to end
- About 1 month
- Delivery interruptions
- 0
Problem
Data and processing workloads had to move between cloud providers without placing every customer at risk at once.
Constraint
Enterprise customers in Australia and the US still depended on daily delivery during the migration.
Decision
Run the target platform in parallel, compare canary outputs, move customers one at a time, and retain a rollback path at every cutover.
System path
Delivery active
- 01Azure production
- 02Canary comparison
- 03Client cutover
- 04GCP production
- Azure
- GCP
- Pub/Sub
- Cloud Storage
- Compute Engine
- GKE
03 · experience
Building and operating software since 2019
Senior Software Engineer · Polybee
Feb 2023 to Present · Singapore
Own customer-facing product, cloud, and data-platform delivery for agricultural analytics across Australia and the US.
- Delivered dashboard visualizations, phenotyping, forecasting, and a linear-programming optimizer from customer feedback and field constraints.
- Operate Prefect workflows processing about 2 TB per week across about 45 daily runs.
- Designed and deployed Django and React services across AWS, Azure, and GCP.
- Improved on-time delivery to 98%, reduced failures to under 5%, and sustained 99.9% uptime.
- Managed interns and mentored 5 to 10 engineers and interns over time.
More production ownership
- Built event-driven AWS ingestion for drone data with missed-event recovery, duplicate protection, and safeguards for incomplete uploads.
- Protected in-flight GPU work while scaling Azure workers from two instances to zero and reduced idle virtual-machine time by about 20%.
- Completed a major cross-cloud migration in about one month using canary runs and client-by-client cutovers.
- Scaled operational reporting to tens of thousands of lots and reduced issue-identification time to under two minutes.
- Delivered browser-based agricultural-row detection matching 118 human-marked rows within seconds while preserving edit, undo, and human-review controls.
- Owned RBAC, least-privilege permissions, and secrets-management infrastructure across three clouds.
- Authored technical documentation and maintained Postman collections for onboarding, handoffs, and troubleshooting.
- Python
- Django
- React
- Prefect
- AWS
- Azure
- GCP
- Kubernetes
Software Development Engineer 2 · OYO Rooms
Jul 2022 to Jan 2023 · Gurugram, India
Built and supported high-traffic hotel-booking services while guiding SDE 1 engineers.
- Built Java and Spring Boot microservices for high-traffic, customer-facing hotel-booking services.
- Supported Kubernetes deployments on Azure.
- Raised automated unit-test coverage by about 30%.
- Guided SDE 1 engineers through implementation questions and delivery work.
- Java
- Spring Boot
- Maven
- Kubernetes
- Azure
Maximl Labs
May 2019 to Jul 2022 · Chennai, India
Progressed from Software Engineer to Senior Software Engineer while building an industrial analytics platform across backend, frontend, and deployment systems.
Senior Software Engineer
Apr 2021 to Jul 2022
- Delivered end-to-end modules for an industrial analytics platform.
- Helped split a monolithic application into independently deployed microservices.
- Built a serverless media-storage service with multi-provider support and pre-signed URLs.
- Expanded analytics APIs and improved mobile reliability in low-connectivity environments.
Software Engineer
May 2019 to Apr 2021
- Built Django, Django REST Framework, Celery, and PostgreSQL services with multilingual and asynchronous workflows.
- Enhanced Angular applications with route guards, reusable npm libraries, HTTP interceptors, and IndexedDB storage.
- Designed data-extraction, analysis, and indexing solutions and documented technical specifications.
- Python
- Django
- Angular
- Celery
- PostgreSQL
- Serverless
- CI/CD
04 · delivery
From customer requirement to production rollout
I work directly with customers, R&D, operations, and company leadership. The job is to understand what must work in the real environment, translate it into a system decision, and stay involved through rollout and support.
Customer path
Track 01- 01
Discover
Understand the requirement and operating context.
- 02
Clarify
Identify constraints, edge cases, and delivery expectations.
- 03
Build
Turn the requirement into product, platform, and data changes.
- 04
Roll out
Release safely for the customer and monitor the result.
- 05
Support
Use field incidents and feedback to guide the next change.
Team path
Track 02- 01
Plan
Break work into clear, reviewable delivery steps.
- 02
Assign
Give engineers and interns the context needed to own a task.
- 03
Review
Lead code reviews, design reviews, and implementation discussions.
- 04
Mentor
Answer questions, unblock delivery, and help people improve.
- 05
Operate
Keep product, pipeline, automation, and observability work aligned.
Leadership proof
Managed interns and mentored 5 to 10 engineers and interns over time.
05 · process
AI-assisted engineering, with human control
I use Claude Code and Codex in daily engineering work to research standards, maintain task context, implement focused changes, and review code from independent perspectives. Rules, validation, and human approval remain part of every change.
01
Clarify
Understand the requirement, risk, and expected result.
02
Research
Check the codebase, external standards, and existing constraints.
03
Design
Prepare the approach and obtain approval before implementation.
04
Implement
Make focused, reviewable changes with persistent task context.
05
Review
Use independent review passes for code, rules, security, and regressions.
06
Validate
Run the relevant checks and verify the actual output.
07
Approve
Keep a human decision at every material change or external action.
The result is faster, more deliberate software delivery. My production experience remains grounded in software systems, cloud platforms, and data pipelines.
06 · capabilities
Tools organised by the systems they support
Languages and application development
- Python
- Java
- JavaScript
- TypeScript
Backend and data
- Django & Django REST Framework
- Spring Boot & Maven
- Celery & REST APIs
- PostgreSQL & Redis
- Indexing, caching & performance tuning
Frontend and customer interfaces
- React & Angular
- Ant Design
- Responsive interfaces
- Data visualization
- Pagination & asynchronous loading
- Browser performance
Cloud and platform
- AWS
- Azure
- GCP
- Docker
- Kubernetes
- Nginx
- Gunicorn
- GitHub Actions
- CI/CD
Provider services
- AWS: S3, Lambda, ECS, ECR, RDS, ALB, CloudFront, Route 53, EC2
- Azure: Service Bus, Functions, Blob Storage, Storage Tables, Virtual Machines, Virtual Machine Scale Sets
- GCP: Pub/Sub, Cloud Storage, Compute Engine, Google Kubernetes Engine
Data pipelines and MLOps infrastructure
- Prefect
- Event-driven architecture
- Idempotency & distributed locks
- Retries, backoff & checkpoints
- Scheduling & operational SLAs
Observability and operations
- Logging, metrics & tracing
- Elastic & Sentry
- Supabase & Metabase
- n8n & Microsoft Teams alerting
- Incident response
Security and documentation
- RBAC & least privilege
- Secrets management
- Passbolt & Tailscale
- Postman
- Technical documentation
AI-assisted engineering
- Claude Code & Codex
- Multi-agent code review
- Persistent task context
- External standards research
- Validation checks
- Human approval
07 · about
Engineering for the part after the demo
I'm Shobhit, a Senior Software Engineer based in Singapore. Over the past 7+ years, I have built customer-facing products, backend services, cloud platforms, and data-processing systems across several industries.
I like engineering that becomes quiet once it works: a dashboard that responds immediately, a pipeline that can recover safely, or a cloud migration that customers do not notice. That usually means understanding the operating context before choosing the technical solution.
I also use AI as part of my daily engineering process. It helps with research, implementation, and independent review, while clear rules, validation, and human approval keep the work grounded.
I'm open to senior software engineering roles in Singapore and internationally, with relocation and travel.
Recommendations
“Shobhit is a remarkable full-stack developer with a knack for architecting software solutions seamlessly across all layers, from frontend to backend and database design.”
“He consistently brings clarity, ownership, and technical depth to the table.”
“Technically sound and highly recommended to any organization. Shobhit is always trying to contribute his best to the team and also finds time to guide others when needed.”
08 · contact
Building software that has to work outside a demo? Let's talk.
I'm open to Senior Software Engineer opportunities in Singapore and internationally. I can relocate and travel for the right role.