Senior Backend Engineer with 5+ years building scalable SaaS platforms, distributed data pipelines, and production ML/LLM systems. I work across Python services, REST and inference APIs, cloud data platforms, GenAI/RAG workflows, MCP, OCR extraction, and resilient enterprise systems.
DAG workflows, async queues, caching layers, and service-bus orchestration for resilient production systems.
Python microservices, REST and inference APIs, Django, FastAPI, and scalable data-service platforms.
Bedrock, RAG, OCR extraction, vector databases, Databricks, Snowflake, and ML-driven workflows.
CI/CD, Kubernetes, observability, release governance, schema safeguards, and reliability-first delivery.
Backend architecture · Distributed data · Applied AI
I design systems, not just services. Most of what I own starts as a question about boundaries: what belongs in a request path and what belongs behind a queue, where state is allowed to live, which failures a caller should ever see. I work mainly in Python and Scala, across distributed data pipelines, service architecture, and the orchestration that holds the whole thing together.
The stack follows from those decisions rather than the other way round. Kubernetes and Docker for the runtime, AWS and Azure underneath, Kafka, SQS and Service Bus wherever work should be asynchronous, and Snowflake, Databricks, PostgreSQL, MongoDB, Redis or Cosmos DB depending on the access pattern. What I actually spend my judgement on is idempotency, caching strategy, back pressure and observability, because those are what decide whether a system survives its second year.
A growing part of that architecture is LLM inference. I treat a model as one more unreliable dependency with a latency budget and a failure mode: batching and caching to keep inference cost predictable, streaming where someone is waiting, fallbacks for when a provider degrades, guardrails on what a model is allowed to return, and evals that run before a prompt change ships rather than after. I build RAG pipelines, document extraction workflows and agentic systems where the model does the fuzzy reasoning and deterministic services own everything that has to be correct. Knowing exactly where to put that boundary is the real engineering work.
Designing scalable APIs, microservices, and robust backend architectures using modern technologies.
Building intelligent systems with machine learning, deep learning, and generative AI technologies.
Implementing cloud-native solutions on AWS and Azure with focus on performance and scalability.
Building a configurable healthcare data pipeline that handles fax, email, and manual intake; runs Gemini OCR and Bedrock-backed extraction; writes editable metadata back into Cosmos DB; and drives downstream DAG stages for benefit verification, prior authorization, provider and patient communication, autonomous calling, and scheduling.
Config-driven referral pipeline from intake sources through orchestration, Gemini OCR, Bedrock extraction, Redis enrichment, Cosmos DB writeback, review UI, benefit verification, prior authorization, provider and patient communication, and scheduling.
Taking ownership of engineering responsibilities across healthcare and senior-living applications, with a focus on data modules, platform workflows, and full-stack systems that support data-driven product capabilities, including referral extraction workflows powered by Bedrock-integrated data extraction systems.
Built production-grade ingestion, orchestration, and agentic AI workflows for healthcare data platforms, focusing on scalable pipelines, governed data operations, and reliable high-throughput processing.
Led SaaS development projects across Python REST frameworks, AWS, big data, containerization, document intelligence, and GenAI workflows for enterprise contract platforms.
Delivered enterprise data-service products using React, JavaScript, Python, SQL, MongoDB, Bash scripting, and Azure, with a focus on scalable client-facing analytics and reporting systems.
Building Bedrock-integrated extraction workflows that transform unstructured referral inputs into structured product data for healthcare operations.
Built resilient ingestion and recommendation workflows with confidence-aware pattern inference, human review routing, and distributed job orchestration.
Extended Snowflake-first data pipelines to Databricks with governed Unity Catalog writes, metastore-aware operations, and strict schema safeguards.
Delivered data-intensive dashboards with 30+ visualizations, interactive drilldowns, optimized queries, and data-quality monitoring for healthcare workflows.
Designed and scaled a production-grade data ingestion, prediction, dashboard, and scorecard system that contributed to ~$70M annual revenue.