Open to Global Contract & Senior Full-Time Roles

Shivam Gupta

Senior Backend & AI Engineer

Backend engineer with 7+ years building secure, multi-tenant architectures and distributed data pipelines. Specializes in AI infrastructure, authentication layers, and migrating heavy compute workloads to Kubernetes to cut infrastructure costs.

Get in Touch View Architecture

GitHub LinkedIn contact@shivamai.dev

Experience

OneCard

Senior Software Engineer

CurrentFeb 2023 — Present

80%

service cost cut

~50%

overall compute cut

Re-architected a core data-transformation pipeline, migrating dbt from AWS Glue to PySpark on Kubernetes (EKS), cutting the service cost by 80% and driving a ~50% reduction in overall infrastructure compute costs.
Built a Python authentication library integrating OAuth2 and AWS Secrets Manager for unified, RBAC-compliant security across sessions and scheduled jobs.
Shifted compute operations to Trino and rewrote Pandas logic in Polars, reducing compute costs by 40% on scaled workloads.
Designed a SQLAlchemy query-translation layer to eliminate SQL injection vulnerabilities and syntax errors.
Deployed a Flask/React monitoring dashboard on Kubernetes (EKS) via ArgoCD for real-time distributed cron visibility.

PythonPySparkPostgreSQLFlaskEKSdbtTrinoPolarsOAuth2ArgoCD

Shiprocket

Software Engineer

Jun 2021 — Feb 2023

95%

manual accounting removed

Automated fulfillment billing with a rules-based decision engine, eliminating 95% of manual accounting workload.
Resolved critical backend performance bottlenecks to maintain platform stability through peak seasonal traffic.

PHPLaravelEloquent ORMMySQLJavaScriptRedis

Featured Systems & Architecture

System 01

LLM Cost & Latency Observability

Built a FastAPI service with Pydantic-validated endpoints wrapping the Gemini API. Implemented retry logic with exponential backoff and a budget-cap circuit breaker to prevent runaway token costs. Engineered prompt routing between two model tiers based on input complexity.

View Repository

Reliability

Exponential backoff retries around upstream model calls.

Cost Control

Budget-cap circuit breaker halts runaway token spend.

Routing

Complexity-based prompt routing across two model tiers.

PythonFastAPIPydanticGemini APISQLitePrometheusGrafana

Technical Competencies

Backend & APIs

PythonFastAPIRESTful APIsPydanticFlaskSQLAlchemySystem Design

Cloud & Infrastructure

AWS EKSEC2S3IAMSecrets ManagerEventBridgeKubernetesDockerArgoCDCI/CD

Data & Distributed Systems

PostgreSQLApache KafkaApache AirflowTrinoApache SparkPolarsRedis

AI Engineering

Agentic WorkflowsGoogle ADKClaude CodeAmazon QGemini API