Senior Python Engineer - GenAI & RAG Systems

TEJA

*********@*****.*** +1-704-***-**** linkedin.com/in/teja-k9409

PROFESSIONAL SUMMARY

●Senior Python Engineer with 10+ years of backend and distributed systems experience, specializing in FastAPI/Flask services, REST API design (OpenAPI), and production AI integration.

●Cut manual document processing effort by approximately 25% by architecting and shipping a production RAG and multi-agent AI pipeline integrating Azure OpenAI, AWS Bedrock, and Google Vertex AI (Gemini API) into FastAPI backend services.

●Stabilized a production AI service under peak load by diagnosing a retrieval latency spike, optimizing query patterns, and adding targeted monitoring, restoring response times through a CI/CD-deployed fix.

●Builds resilient, secure REST and event-driven services: implements caching, rate limiting, retries, and circuit breakers, and secures APIs with OAuth 2.0, OIDC, and JWT-based service-to-service authentication.

●Delivers full stack solutions pairing Python/FastAPI and Java Spring Boot backends with React and Angular frontends, backed by PostgreSQL/MongoDB and Redis caching, deployed on AWS and GCP with Docker, Kubernetes, and Terraform.

●AWS Certified Solutions Architect - Associate with a Master of Science in Computer Science from Texas A&M University-Kingsville.

TECHNICAL SKILLS

GenAI, LLM & Agentic AI: Azure OpenAI, AWS Bedrock, Google Vertex AI (Gemini API), LangChain, LangGraph, AutoGen, Semantic Kernel, RAG, Multi-Agent Orchestration, Tool Calling, Agent State Management, Prompt Engineering, Embeddings, Sentence Transformers, MCP, Agent2Agent (A2A) Protocol, Azure AI Foundry, Azure AI Search, Azure ML

LLMOps & Evaluation: LangSmith, Ragas, Model/Prompt Evaluation Workflows, Regression Testing, Model Version Governance, Streamlit (internal tooling), Darwin (internal AI model inventory)

Vector & Retrieval: Pinecone, ChromaDB, FAISS, Milvus, Weaviate, Semantic Search, Hybrid Retrieval, Document Retrieval

API Design & Integration: REST APIs, OpenAPI Specification, Webhooks, Event-Driven Services, Microservices

Machine Learning & Deep Learning: TensorFlow, PyTorch, Keras, Hugging Face Transformers, Scikit-learn, MLflow, Model Fine-Tuning and Training, Kubeflow (evaluated)

Programming Languages: Python, Java, JavaScript, TypeScript, SQL, C#, Rust

Backend & Full Stack: FastAPI, Flask, Spring Boot, Spring MVC, Spring Batch, Node.js/TypeScript, REST APIs, Microservices, Pydantic, Alembic

Frontend: React.js, Angular (Angular 15+, NgRx, RxJS, Signals), TypeScript, JavaScript, HTML5, CSS3

Cloud & Infrastructure: AWS (EC2, S3, Lambda, SQS, ECS, EKS, SDK v3), Azure (Container Apps, AKS, Functions, API Management, Event Grid, Service Bus, SQL, AD B2C), GCP (Cloud Run, AlloyDB, API Gateway, Cloud Storage, Cloud Functions), Docker, Kubernetes, OpenShift, Terraform, SaltStack, Service Mesh, Multi-Tenant/BYOC Architecture

AI Reliability & Security: Retry/Fallback Handling, Circuit Breakers, Guardrails, Secure Tool Execution, Data Isolation, PII-Aware Access Controls

Reliability & Production Operations: SLO/SLA Management, Disaster Recovery, Resiliency Patterns, Autoscaling, Load Testing, Latency Optimization, Failover Routing, Caching, Rate Limiting

Model Serving & Inference: NVIDIA Triton Inference Server, KServe, Seldon, Ray Serve, GPU-based Inference

Data Engineering: Databricks, Apache Spark/PySpark, AWS Glue, Pandas

CI/CD & DevOps: Jenkins, GitHub Actions, Azure DevOps, Git, Bitbucket, JFrog Artifactory, GitHub Copilot, Build Automation

Databases: PostgreSQL, Oracle, MongoDB, Cosmos DB, Redis, NoSQL, Data Modeling

Messaging & Integration: Apache Kafka, AWS SQS, Azure Service Bus, Event-Driven Architecture, Service Integration

Testing & Quality: PyTest, JUnit, Mockito, Jest, React Testing Library, Jasmine, Karate, Playwright, Integration Testing, API Testing, Test-Driven Development

Observability: Datadog, Splunk, Grafana, New Relic

Security & Access Control: OAuth 2.0, OIDC, JWT, RBAC, SSO, SAML, MFA, HashiCorp Vault, AWS Secrets Manager, Azure Key Vault

Contact Center Integration: Amazon Connect, Genesys

Agile & Collaboration: Agile/Scrum, Sprint Planning, Technical Design Reviews, Code Reviews, Mentoring, Cross-functional Collaboration, Stakeholder Communication

PROFESSIONAL EXPERIENCE

Zyme Works Middletown, DE Jan 2024 - Present

Senior Software Engineer - Enterprise AI and Application Modernization Platform

●Reduced manual document processing and review effort by approximately 25% by architecting and shipping an automated RAG-powered document review pipeline in Python and FastAPI, serving as the team's primary GenAI engineer within a 6-10 engineer team.

●Kept an enterprise RAG system stable at production scale (thousands of documents per day, mixed batch and real-time query traffic) by building latency, volume, and failure-rate monitoring and demand-based autoscaling into the processing layer.

●Restored response times and stabilized the system under peak traffic by diagnosing a production latency spike through the FastAPI service, retrieval, and LLM call layers, root-causing it to an inefficient retrieval query pattern, and shipping an optimization and monitoring fix through CI/CD.

●Enabled enterprise customers to run workloads in their own cloud environments on one shared codebase by designing the platform's service layer for multi-tenant isolation with Bring Your Own Cloud (BYOC) deployment support.

●Let AI-heavy services scale and deploy independently from the rest of the platform by architecting domain-decomposed Python FastAPI microservices exposing REST and OpenAPI-documented endpoints, consumed by React, Angular, and TypeScript frontends.

●Cut manual handoffs in document processing by designing multi-agent workflows in AutoGen and the Agent2Agent protocol with specialized agent roles and defined tool-calling contracts against internal APIs, piloting Semantic Kernel for select orchestration scenarios.

●Enabled agents to plan multi-step actions and recover mid-workflow by using LangGraph's stateful graph execution to manage agent state and control flow, instead of chaining single-shot LLM calls.

●Improved enterprise information-retrieval efficiency by approximately 30% by building RAG pipelines in LangChain and Azure OpenAI with chunking, metadata filtering, and Pinecone/ChromaDB/FAISS vector search, evaluating Milvus and Weaviate for hybrid retrieval at greater scale.

●Cut new-tool onboarding time for agent workflows by standing up Model Context Protocol servers giving LLM agents a consistent, permissioned, and authorization-scoped way to reach internal APIs and data sources.

●Prevented a single failed tool call from breaking an entire agent workflow by implementing retry and fallback guardrails around agent tool calls, with graceful degradation on malformed or failed invocations.

●Let existing enterprise components consume LLM-powered retrieval without a platform rewrite by extending a Java Spring Boot service layer to integrate with the Python AI services over REST.

●Surfaced AI-generated document insights directly to contact center agents by integrating the platform with Amazon Connect and Genesys, tied into event-driven workflow status updates via webhooks and messaging.

●Cut deployment and release effort by approximately 25% by building Jenkins and GitHub Actions CI/CD pipelines with JFrog Artifactory-managed build artifacts and container images, provisioning AWS EC2/S3/Lambda/SQS via AWS SDK v3 and evaluating AWS Bedrock and Google Vertex AI (Gemini API) as alternate model providers.

●Reduced asynchronous workflow processing time by approximately 20% by introducing Kafka-based event streaming for document ingestion and processing status updates, combined with AWS SQS to decouple upstream and downstream services.

●Protected platform availability and retrieval latency under peak enterprise usage by defining SLO/SLA targets, configuring autoscaling, adding caching and rate limiting, and running load testing backed by health checks, circuit breakers, and failover routing as part of the platform's disaster recovery design.

●Secured enterprise data and AI-driven actions within authorized boundaries by implementing OAuth 2.0, OIDC, JWT, RBAC, and SSO based access controls scoped per tenant across services and agent tool access.

●Cut time to diagnose production issues by using Datadog and Grafana to trace latency spikes and failed retrieval calls back to specific pipeline stages, and by writing PyTest/JUnit unit and integration tests covering FastAPI services and RAG pipeline components.

●Standardized how new LLM versions reached production by setting up Azure AI Foundry regression-testing workflows and an evaluation loop (LangSmith, Ragas) scoring retrieval and response quality on every meaningful pipeline change, tracked in the internal Darwin model inventory for governance.

●Accelerated team development velocity by mentoring junior engineers on Python and Spring Boot service design and debugging, and introducing GitHub Copilot alongside existing code review practices.

●Wrote architecture documentation, API contracts, and deployment runbooks that new team members used to ramp up on the platform.

●Took part in an on call rotation, triaging and resolving production incidents across the AI platform and its supporting services.

●Partnered with architects, QA engineers, data teams, and business stakeholders to scope features, review technical designs, and validate releases before rollout.

Elevance Health Virginia Jul 2020 - Dec 2023

Full Stack Developer - Enterprise Healthcare Financial Analytics and Claims Applications

●Gave financial analysts a single consolidated dashboard, replacing manual checks across several internal systems, by delivering React and TypeScript interfaces backed by Java Spring Boot REST APIs.

●Reduced manual data validation and transformation effort by approximately 20% by building Python services for claims data validation, transformation, and reconciliation, replacing manual spreadsheet-based checks.

●Reduced recurring operational processing effort by approximately 20% by writing Python automation scripts for reconciliation and reporting tasks previously run by hand.

●Improved information-retrieval relevance by approximately 25% over exact keyword matching by applying Sentence Transformers and embeddings to add semantic search across claims and financial documents.

●Improved reporting job efficiency against high-volume claims data by designing targeted database schema updates and query optimizations.

●Cut manual review of malformed records by building internal utilities to parse and validate semi-structured claims files, including SOAP/WSDL based integrations with legacy systems.

●Caught defects before production releases by validating Java and Python service changes through JUnit and PyTest based unit, integration, and API tests.

●Kept local and production runtime behavior consistent across Azure hosted environments by configuring Jenkins CI/CD pipelines and containerizing components with Docker.

●Reduced time to resolve service dependency issues by tracing production incidents through application logs, database queries, and API responses using Splunk, Grafana, and New Relic.

●Applied HIPAA and GDPR aligned security and access control practices while working with sensitive healthcare and financial application data.

●Supported investment analyst and portfolio dashboard tooling within the financial research repository, building reporting views for capital markets stakeholders.

●Reviewed pull requests and helped enforce coding standards across the team, catching design and defect issues before they reached QA.

●Took part in sprint planning and backlog grooming, helping scope technical work against business priorities for claims platform features.

●Collaborated with business analysts, data engineers, and QA teams throughout planning, development, and release cycles for claims platform enhancements.

CtrlS Hyderabad, India Feb 2017 - Dec 2018

Junior Full Stack Developer - Enterprise Java Full-Stack Applications

●Operations workflows were being tracked manually, so developed Java 8 and Spring Boot services and REST APIs and paired them with React and JavaScript interfaces, giving operations staff a real-time web based view into work that had previously lived in spreadsheets and email.

●Refactored tightly coupled application modules into independently deployable REST based services, including select modules deployed on IBM WebSphere, making it possible to troubleshoot and release components without redeploying the whole application.

●Designed SQL queries and relational data access logic supporting transactional processing and reporting for the operations platform, including integrations with mainframe batch-processing data flows.

●Wrote performance-sensitive backend utilities in Rust that integrated with the Java/Spring Boot service layer for select high-throughput processing tasks.

●Set up Jenkins based build and deployment automation, standardizing how code moved from development into test and production environments and reducing deployment effort by approximately 20%.

●Wrote unit and integration tests covering backend services and API behavior to catch defects before releases, and built Python utilities and automated testing for recurring development and validation tasks, reducing that repetitive effort by approximately 15%.

●Worked with GCP Cloud Storage and Cloud Functions for select file storage and lightweight serverless processing needs.

●Built internal tooling for log parsing and error triage, making it faster to trace recurring application issues back to a root cause during production support.

●Worked with QA and development teams to investigate and resolve defects identified during testing and production support.

●Took part in code reviews and pair programming sessions that helped establish consistent coding practices across the full stack team.

●Supported database schema changes and data migrations as application requirements evolved over the project lifecycle.

Code DHA Hyderabad, India Jan 2015 - Jan 2017

Associate Software Engineer - Enterprise Java Web Applications

●Developed Java based web applications implementing business logic and server side functionality for internal enterprise workflows, translating requirements gathered directly from business stakeholders into working features.

●Built and consumed REST APIs connecting frontend JavaScript components with backend Java services, including select integrations involving FHIR and HL7 data formats for healthcare-adjacent workflows.

●Implemented SQL queries and stored procedures supporting core application data access and reporting needs.

●Performed unit testing, debugging, and defect analysis throughout development and release cycles, laying the API- and database-level testing habits carried into later roles.

●Supported production troubleshooting and post release maintenance for deployed enterprise applications.

EDUCATION

Master of Science in Computer Science

Texas A&M University-Kingsville Jan 2019 - Jun 2020

Bachelor of Technology in Information Technology

Anil Neerukonda Institute of Technology and Sciences (ANITS) Apr 2015

CERTIFICATION

AWS Certified Solutions Architect - Associate

Contact this candidate