Available for senior AI roles

Aayog
Koirala

Senior AI & GenAI Engineer

I build production AI systems in regulated industries — from RAG pipelines and agentic workflows to enterprise-scale document intelligence. Banking. Healthcare. Insurance. The hard stuff.

01 / About

Building AI that works
where it has to.

I'm a Senior AI and GenAI Engineer with 5+ years building production AI systems across financial services, healthcare, and insurance. My work lives at the intersection of LLM architectures, data engineering, and regulated deployment — where "it works in a notebook" means absolutely nothing.

I've designed RAG pipelines that serve sub-second financial document retrieval, built agentic workflows using LangGraph and MCP for multi-step document auditing, and shipped LLM-backed systems that operate under compliance constraints most engineers never have to think about. I'm a backend-first engineer: Python, FastAPI, Spark, and whatever infrastructure the problem demands.

Before GenAI was a buzzword, I was building data pipelines, classical ML models, and NLP systems. That foundation is why my LLM systems actually work in production — I know what breaks, what scales, and what auditors ask about.

5+
Years in Production AI
4
Regulated Industries
3
Cloud Platforms (Azure, AWS, GCP)
02 / Experience

Career Timeline

Senior AI and GenAI Engineer

Current
Western Alliance BankPhoenix, AZ (Remote)
Feb 2024 — Present

Leading delivery of production-grade AI capabilities for finance-domain workflows. Architecting LLM-backed document processing, RAG retrieval, and agentic automation under strict regulatory constraints.

  • Designed end-to-end LLM systems for financial document ingestion, classification, extraction, and summarization under compliance constraints
  • Architected RAG pipelines with Weaviate achieving sub-second retrieval of financial policies and historical records via hybrid dense vector + metadata filtering
  • Built agentic AI workflows using LangGraph and MCP for stateful multi-step document auditing with standardized tool integration
  • Deployed intelligent routing logic for financial transaction approvals, significantly reducing manual review volume
  • Integrated DeepEval quality gates into CI/CD, enforcing faithfulness and relevancy checks on LLM outputs before release
  • Owned architecture decisions balancing model accuracy, latency, cost ceilings, and regulatory expectations
Azure OpenAIAWS BedrockLangGraphMCPWeaviateFastAPISparkDatabricksMLflowDeepEvalDockerKubernetes

AI Engineer

C3 AIRedwood City, CA
May 2022 — Feb 2024

Applied AI and data engineering role focused on Python-driven data foundations and backend workflows, transitioning into LLM evaluation and early generative AI initiatives.

  • Engineered scalable Python services for processing and enriching large structured/semi-structured datasets for production AI workloads
  • Led controlled experiments benchmarking GPT-style models against legacy NLP systems with custom quantitative comparison tooling
  • Developed prompt structuring strategies to stabilize model outputs and reduce hallucinations during experimental LLM phase
  • Supported NLP-based text processing for classification, summarization, and information extraction
  • Containerized key workflows for consistent execution across environments
PythonNLPGPT ModelsPrompt EngineeringCI/CDDocker

Data & Applied Analytics Engineer

Ardent HealthNashville, TN
Jul 2021 — May 2022

Built data pipelines and applied analytics in healthcare, processing sensitive records under strict data handling requirements with early ML and NLP initiatives.

  • Built data ingestion and transformation pipelines for sensitive structured and semi-structured healthcare records
  • Implemented rule-based and classical NLP for document tagging, keyword extraction, and text classification
  • Designed data models and quality checks supporting analytics, reporting, and early AI-adjacent workflows
  • Enforced deterministic behavior and reproducibility due to regulatory sensitivity
  • Integrated pipelines with backend systems while enforcing access controls and audit expectations
PythonNLPData PipelinesDockerCI/CDData Governance

Software and Data Engineer

Westfield InsuranceWestfield Center, OH
Jun 2020 — Jul 2021

Backend and data-focused engineering supporting data-intensive applications, ETL workflows, and operational systems in insurance.

  • Developed Python services for processing, validating, and transforming structured datasets for reporting and operational systems
  • Designed and maintained ETL workflows ingesting data from multiple upstream sources
  • Built and optimized SQL queries and data models for analytics and reconciliation
  • Implemented validation checks and logging for upstream data quality monitoring
PythonSQLETLData PipelinesBackend Services
03 / Skills

Technical Expertise

A production-hardened stack built across regulated industries — not a list of tutorials completed.

Applied AI & Generative AI

LLM IntegrationLangGraphModel Context Protocol (MCP)Prompt EngineeringRAG PipelinesEmbeddings & Vector SearchFine-TuningAgentic WorkflowsDocument Intelligence

Machine Learning & NLP

PyTorchScikit-learnSpacyFeature EngineeringSupervised/Unsupervised ModelingClassical ML BaselinesA/B TestingCross-Validation

Programming & Backend

PythonPostgreSQLSQLFastAPIREST APIsAsync ProcessingService Orchestration

Data Engineering & Pipelines

SparkDatabricksKafkaAirflowBatch & Near-Real-Time PipelinesETL

MLOps & Deployment

CI/CD for AI/GenAIMLflowDockerKubernetesCanary ReleasesPrometheusGrafanaELKOpenTelemetry

Cloud & Infrastructure

Azure OpenAIAzure MLAWS BedrockAWS Compute & StorageGCPTerraformARM TemplatesCDK

Datastores

PostgreSQLWeaviatepgvectorRedisNoSQL

Security & Compliance

IAMSecrets ManagementEncryptionAudit LoggingSecure API DesignDevSecOpsCompliance-Aware AI Delivery
04 / Education & Research

Academic Background

Education

Master of Science in Computer Science
Miami University
Bachelor of Science in Computer Science
Miami University

Certifications

Microsoft Certified: Azure AI Engineer Associate
AI-102
ICAgile Certified Professional
ICP
Master's Thesis

360° Panoramic View Synthesis via Depth-Based Multiplane Images

Engineered a PyTorch framework for spherical view synthesis by constructing Multiplane Images from cube map projections and metric depth estimation (DepthAnything v2). Achieved comparable PSNR/SSIM scores versus a Google Research neural network MPI baseline, providing an efficient, interpretable alternative for VR/AR applications.

PyTorchDepthAnything v2Multiplane ImagesPSNR/SSIMVR/ARComputer Vision
05 / Contact

Let's Build
Something.

I'm open to senior AI engineering roles, technical advisory, and interesting problems in regulated AI. If you're working on something that needs production rigor, let's talk.