Principal Architecture · AI & Agentic Systems · Cloud at Scale

Leif Taylor

Principal engineer shipping revenue-critical AI, cloud, and distributed systems — using agentic workflows to deliver faster, more securely, and with measurable business outcomes.

Twenty years across DevOps, distributed systems, and AI strategy. Currently Principal Engineer and Director of DevOps at ConnectBase. Past lives at Mobile Heartbeat, Actifio, and a conservatory.

Download résumé (PDF)LinkedIn

NOWShipping the direct-to-supplier ordering platform at ConnectBase · Spearheading agentic dev workflows in production

  1. 01$600K+Annualized Azure spend reduction at ConnectBase — ~30% of total cloud cost — through resource right-sizing, workload consolidation, and configuration discipline.
  2. 0230+Legacy VM-hosted services lifted into stateless, horizontally-scalable microservices — pricing, ordering, ingestion, and the supporting platform spine.
  3. 0310K+ / wkSupplier products ingested, normalized, and embedded for semantic retrieval through an LLM-driven pipeline that runs on its own.
  4. 040 → 1Direct-to-Supplier Ordering platform shipped greenfield — wholesalers and channel partners now purchase from thousands of suppliers in one interface.

Architecting systems on the critical path of revenue, delivery, and scale.

Leif Taylor is a principal software engineer and technical director focused on production systems where correctness, operability, and business impact have to be designed together — distributed platforms, AI-enabled workflows, cloud infrastructure, and data-intensive B2B applications.

His current work centers on agentic delivery: pairing the depth of twenty years on the platform side with the leverage of AI-augmented authorship, evaluation harnesses, and tool-using agents to ship secure, observable systems faster than a traditional team profile would predict.

Selected architecture & delivery work.

A handful of named systems with the operational and business problems they solved. Implementation details are in the experience section below.

  • ConnectBase

    Distributed Pricing Engine

    Reengineered the monolithic pricing engine into a worker-driven microservice mesh.

    • Concurrent quote computation across heterogeneous supplier tariffs
    • Fault-isolated worker model — bad inputs no longer take down the engine
    • Horizontal scale-out path the legacy engine couldn’t support
    • Distributed workers
    • Pricing
    • Stateless
    • Scalability

  • ConnectBase

    Direct-to-Supplier Ordering

    Chief developer and owner of the new ordering platform connecting wholesalers and channel partners to thousands of suppliers in one interface.

    • 0→1 product, greenfield architecture
    • Multi-supplier order routing, normalization, and reconciliation
    • Foundation for the next generation of ConnectBase commerce flows
    • Greenfield
    • B2B Commerce
    • Multi-tenant
    • Domain modeling

  • ConnectBase

    AI-Driven Supplier Ingestion

    LLM-driven catalog ingestion, semantic search (RAG), and TAM analysis on the data layer.

    • 10K+ supplier products processed weekly with no human in the loop on the happy path
    • Semantic knowledge-base search across the entire ingested corpus
    • Foundation for downstream intelligence: TAM, gap analysis, recommendations
    • RAG
    • Semantic search
    • LLM workflows
    • Data ingestion

  • ConnectBase

    Cloud Cost Program

    FinOps-style discipline across Azure: resource right-sizing, environment consolidation, and configuration hygiene.

    • $50K+ monthly spend reduction
    • $600K annualized savings
    • ~30% reduction in total cloud expenditure
    • Azure
    • FinOps
    • Operational discipline

Twenty years from infrastructure into principal architecture.

A progression from automation and DevOps leadership into product-critical systems ownership, AI strategy, and end-to-end delivery.

  1. 2023 — Present
    CURRENT

    ConnectBase

    Principal Engineer

    • Re-engineered the Pricing Engine into a distributed, worker-driven microservice system — horizontal scalability, fault isolation, concurrent quote computation
    • Chief developer and owner of the new Direct-to-Supplier Ordering solution — a 0→1 product letting wholesalers and channel partners purchase from thousands of suppliers in one interface
    • Spearheading AI strategy: LLM-driven supplier ingestion (10K+ products/week), RAG knowledge base, TAM analysis, and adoption of agentic dev workflows in production
    • Refactored legacy raw-SQL data paths into a type-safe DAL/ORM abstraction — query standardization, injection-risk reduction, cross-service

    Promoted from

    ConnectBase

    Director of DevOps

    2023 — 2025

    • Modernized 30+ legacy VM-hosted services into stateless, horizontally-scalable microservices
    • $600K+ annualized Azure savings through resource and configuration optimization
    • Stood up enterprise-wide feature-flag, secrets-management, and progressive-delivery (canary) infrastructure
    • Built CI/CD pipelines for 30+ services across Java, TypeScript / Node, Python, and C#
    • Stood up unit / functional / e2e coverage across the product suite using AI-assisted test authorship
    • Disaster-recovery and outage point person
  2. 2019 — 2023

    Mobile Heartbeat

    Principal DevOps Lead

    Promoted from Senior DevOps Engineer

    • Designed a serverless internal application testing framework, fully Terraform-driven, integrated into CI/CD
    • Built CI/CD for multitenant microservice apps on Azure — Terraform + Flux/Helm + Kubernetes + automated test deployment + DB bootstrapping
    • Designed CI/CD pipelines for iOS and Android mobile applications
    • Customer-release support across Android, iOS, and Windows platforms
  3. 2016 — 2019

    Actifio

    Head of DevOps

    Promoted from DevOps Intern → Lead → Head

    • Chief developer of the automated regression testing framework — Python/Node + Docker + Ansible + Jenkins + Robot Framework
    • Managed 100+ dev databases (Oracle, SQL Server) and 2,000+ VMs across on-prem ESXi, Hyper-V, physical hosts, and AWS EC2
    • Travelled to India to mentor and code-review the distributed engineering team
    • Built developer-portal tooling: test execution, host management, log analysis
  4. 2009 — 2015

    Professional Music

    Pianist · Conductor · Choral Director

    • A conservatory-trained background in composition, theory, and performance — informs a systems approach to structure, constraint, coordination, and execution under pressure.
  5. 2005 — 2009

    San Francisco Conservatory of Music

    B.M. — Composition, Music Theory, and Piano

Founder-stage and enterprise-facing engagements.

Architectural and technical-director work alongside the principal day-job — across AI, marketplace, scientific, and creative software domains.

  • 2025 — Present

    Chemveric

    Technical Director

    Architected and led end-to-end development of a B2B SaaS cheminformatics marketplace connecting buyers and suppliers of research chemicals and materials. Multi-tenant NestJS / Postgres / RDKit / AWS, with AI-assisted ingestion, RFQ workflows, and funding-opportunity matching.

    • NestJS
    • PostgreSQL
    • RDKit
    • AWS
    • Marketplace

  • 2022 — Present

    Alacrity Solutions

    Founding Member · Principal Architect

    AI consulting practice helping F500s operationalize LLMs for software development, customer support, finance, compliance, and scientific data. Delivered work for global tax-tech, computational-science, and industrial-chemistry clients.

    • RAG
    • Compliance
    • Scientific data
    • LLMs

  • 2022 — 2023

    Imprint.live

    CTO · Principal Architect

    Led development of an AI-driven social platform — OpenAI-powered moderation, synthetic-user systems, user-value analytics for stakeholders.

    • OpenAI
    • Moderation
    • Social platform

  • 2020 — 2024

    Open Interpreter & Storymill AI

    Contributor · Founding Member

    Contributed to local-LLM execution patterns for Open Interpreter; founding member and chief architect of Storymill AI, an AI-assisted long-form creative tool for novels, screenplays, and short fiction.

    • Local LLMs
    • Agentic tooling
    • Creative AI
06AI & Agentic systems

Production AI, not demonstrations.

Two decades on the platform side, plus three years building production LLM systems and shipping with agentic dev workflows daily. The work is reliability, eval, observability, cost control, and integration with real business operations — not prompt-engineering theatre.

Systems I build

  • Production RAG with hybrid retrieval
  • Tool-using agents with structured outputs
  • Multi-agent / agentic workflow orchestration
  • Long-running stateful agent processes
  • LLM evaluation harnesses & regression suites
  • Inference-cost & latency optimization
  • AI observability & guardrails

Practices I bring

  • Agentic dev workflows (Claude Code, MCP, sub-agent orchestration)
  • AI-augmented test authorship at scale
  • Context engineering for long-context models
  • Model-Context-Protocol (MCP) tool servers
  • AI-assisted incident response & post-mortem authorship
  • AI-augmented code review and architecture critique
  • Prompt-system version control & eval-driven iteration

Grouped by architectural concern.

Languages, platforms, and tools used to ship and operate production systems across the roles above.

Languages & runtimes
  • TypeScript / Node
  • Python
  • Java
  • C#
  • Objective-C
Frontend & mobile
  • React
  • React Native
  • Angular
  • Svelte
  • Vite
AI / LLM
  • Anthropic
  • OpenAI
  • LangChain
  • MCP
  • RAG
  • pgvector
  • Eval harnesses
Cloud & delivery
  • Azure
  • AWS
  • Cloudflare
  • Kubernetes
  • Terraform
  • Docker
  • Helm
  • Flux
  • GitHub Actions
  • CircleCI
  • Ansible
Data systems
  • PostgreSQL
  • RDKit
  • PostGIS
  • pgvector
  • MongoDB
  • Cosmos DB
  • Vitess
  • Snowflake
  • MySQL
  • Oracle
  • SQL Server
  • Elasticsearch
Observability & ops
  • Datadog
  • OpenTelemetry
  • Grafana
  • Jira
  • Confluence
  • Bitbucket
  • Bamboo

Available for principal architecture, AI systems strategy, and fractional technical leadership.

Greater Boston Area · Open to remote and hybrid engagements. Best for complex cloud, AI, product, and delivery work where architecture and business outcomes converge.

Contact

Code samples available on request — most repositories are private under client NDAs.