Company Overview :
Podstack is a high-growth technology startup focused on building scalable infrastructure for the next generation of AI-driven applications. Operating at the intersection of cloud computing and machine learning, the company provides robust backend solutions that enable developers to deploy and manage complex models with ease. Based in Bangalore, Podstack fosters a culture of technical excellence, rapid iteration, and deep engineering ownership, serving a growing base of enterprise clients looking to optimize their computational workflows.
Role Overview :
As a Backend Engineer at Podstack, you will be responsible for architecting and maintaining the core services that power our platform. You will work closely with the founding team to translate complex technical requirements into high-performance, production-ready code. This role involves managing the lifecycle of our backend systems, from designing efficient APIs to optimising cloud infrastructure for heavy machine learning workloads.
You will have a direct impact on the reliability and scalability of our platform, ensuring that our users can leverage GPU-intensive resources seamlessly.
Key Responsibilities :
1. Own the GPU-hour metering pipeline :
- Capture real-time usage across fractional GPU slices (PodVirt, 12.5% - 100%) and across NVIDIA (CUDA), AMD (ROCm/HIP), and Google (TPU) hardware on a single control plane.
2. Build the rating engine behind our 5-tier pricing model :
- Turning metered events into priced line items with per-tier, per-SKU, and per-route rules.
3. Meter every consumption model :
- Per-GPU-hour (datacenter), per-second (GPU-over-IP), per-cell (serverless notebooks), per-event (serverless stream processing), per-TB-scanned (Data Lake serverless SQL), and managed inference.
4. Ship INR billing and GST-compliant invoicing :
- Plus the metering APIs and INR payment flow behind the self-serve usage portal that powers our product-led-growth motion. Extend to USD billing (Stripe + wire) ahead of the North America launch.
5. Guarantee correctness :
- Idempotent event processing, reconciliation, audit trails, and zero-revenue-leak metering, all under DPDPA data-handling requirements. Integrate everywhere - partner with the control-plane, federated scheduler, and observability teams to wire metering into every billable surface.
Required Skillset :
- Demonstrated proficiency in building backend systems using Golang or Python, with a strong grasp of data structures and algorithms.
- Proven ability to design and manage relational databases like MySQL, ensuring data integrity and query optimization in high-traffic environments.
- Hands-on experience with cloud infrastructure management on AWS or Google Cloud Platform, specifically regarding compute and storage services.
- Strong understanding of container orchestration tools such as Docker and Kubernetes for managing microservices at scale.
- Familiarity with the lifecycle of machine learning models and the specific infrastructure requirements for GPU-based workloads.
- Excellent problem-solving skills and the ability to communicate complex technical concepts clearly to team members in a fast-paced, hybrid work environment.
- A degree in Computer Science or a related field from a reputable institution, reflecting a strong foundation in system design and software engineering principles.
Nice to have :
- Usage-based / consumption pricing experience at a cloud or SaaS provider.
- Kafka / Flink or other streaming infrastructure (relevant to event-level metering).
- Exposure to GPU / ML infrastructure or cloud cost & billing systems.
- Familiarity with Indian GST e-invoicing.
Podstack