Architecting high-throughput backend systems that scale.

Lead backend architecture for multi-national solar forecast blending pipelines, ingesting satellite feeds and ML predictions to calculate solar generation across UK, India, and EU grids.

• Architect and maintain the solar forecast blending service end-to-end across three country deployments, blending day-ahead and 6-hour models with automated failover during satellite telemetry dropouts.
• Refactored the blending engine to be dynamically driven by a centralized BlendConfig rather than hardcoded per-country forks, enabling instantaneous multi-region rollouts via config commits.
• Built Hexatron, an internal operational management portal in Go that triggers Airflow DAGs and surfaces live pipeline execution logs to engineers in real time.
• Migrated 8 forecasting applications onto the internal gRPC Data Platform, cutting data-saving costs 30% by shortening Airflow DAG runtimes.
• Overhauled API authorization to dp-sdk v0.32.0, moving permissions from per-site access to multi-tenant organization scope using HubSpot company metadata in JWT claims.
• Engineered an interactive Streamlit data-science console for machine learning researchers to visualize multi-model forecast divergence.
• Authored 68+ merged pull requests across 10 open-source repositories in the Open Climate Fix org (solar-consumer, quartz-api, pv-site-api, india-forecast-app, uk-pv-forecast-blend, pv-site-production).

Architected core diagnostic logistics and financial transaction pipelines using Go and Python, ensuring high reliability for time-critical medical diagnostics.

• Engineered a decoupled payout architecture in Go backed by AWS SQS and background worker pools, slashing partner payout failure rates by 98% with automated idempotent retries.
• Implemented fine-grained JWT RBAC for partner invoice systems, securing sensitive healthcare billing records across laboratory franchises.
• Built an automated monthly invoicing engine with pre-settlement 'dry run' validation in Go, cutting invoice errors 95%.
• Migrated high-throughput order and sample lookup services from AWS RDS to Elasticsearch, dropping search query latency by 96% while removing database CPU spikes.
• Designed asynchronous task queues and messaging topologies using RabbitMQ, Redis, and AWS SQS for sample collection routing.

Built high-throughput communication microservices and fault-tolerant conversational bot engines processing thousands of requests per second.

• Built the FastAPI chatbot service to handle thousands of requests per second, with session state cached in Redis ElastiCache, cutting latency 33% from 1.5s to under 1s.
• Engineered real-time log ingestion microservices with RabbitMQ streaming to MongoDB, moving writes off the request path to cut request latency 20%.
• Optimized MongoDB data schemas and implemented database sharding, improving complex document lookup speeds by 18%.
• Owned the chatbot availability redesign, choosing an active-active multi-region AWS topology with cross-region replication and failover to end single-region downtime.
• Built an intelligent request retry layer that rescued 5% of peak-traffic dropped webhook payloads.

• Built 4 full-stack applications utilizing the MERN stack (MongoDB, Express.js, React, Node.js) for internal customer support automation.
• Optimized relational and document database query execution plans, delivering a 20% latency reduction.
• Integrated third-party REST APIs and automated cloud deployments, improving test pipeline reliability by 30%.