Embedded Files
Creator of PreMortem AI — a modular, schema-driven LLM system for automated risk discovery,
Featured Project - PreMortem AI
Featured Project - PreMortem AI
AI-powered risk analysis system for production workflows
AI-powered risk analysis system for production workflows
Tech Stack: Python • GPT-5.2 • Pipedream • Make.com
Tech Stack: Python • GPT-5.2 • Pipedream • Make.com
Core Capabilities:
• Turns free-form project inputs into structured, auditable risk reports
• Deterministic LLM pipelines with strict schema validation
• Turns free-form project inputs into structured, auditable risk reports
• Deterministic LLM pipelines with strict schema validation
• Multi-stage orchestration with retries, validation boundaries, and observability
• Multi-stage orchestration with retries, validation boundaries, and observability
→ View Full Case Study
Architecture & Engineering Philosophy
Architecture & Engineering Philosophy
I design AI-powered automation systems with a focus on reliability, modularity, and predictable behavior in production environments. My work emphasizes building AI pipelines that behave consistently, are easy to maintain, and integrate into real engineering workflows.
Rather than optimizing for demos, I optimize for systems that can be trusted, extended, and operated over time.
Core Principles
Core Principles
• Modular, Layered Design
Pipelines are composed of isolated stages (intake, validation, inference, scoring, orchestration, and reporting) to simplify debugging, upgrades, and long-term maintainability.
• Modular, Layered Design
Pipelines are composed of isolated stages (intake, validation, inference, scoring, orchestration, and reporting) to simplify debugging, upgrades, and long-term maintainability.
• Deterministic AI Outputs
LLMs are constrained through schema-first design and structured prompting to produce consistent, machine-usable outputs for downstream automation.
• Deterministic AI Outputs
LLMs are constrained through schema-first design and structured prompting to produce consistent, machine-usable outputs for downstream automation.
• Reliability by Default
Systems include retries, validation gates, rate-limits, and error isolation to ensure stable end-to-end execution under real-world conditions.
• Reliability by Default
Systems include retries, validation gates, rate-limits, and error isolation to ensure stable end-to-end execution under real-world conditions.
• Composable Agent Workflows
Complex tasks are decomposed into focused agents (discovery, scoring, mitigation, and summarization) to improve accuracy and extensibility.
• Composable Agent Workflows
Complex tasks are decomposed into focused agents (discovery, scoring, mitigation, and summarization) to improve accuracy and extensibility.
• Human-Readable Results
Outputs are designed for clarity — structured insights and summaries that support engineering, product, and leadership decision-making.
• Human-Readable Results
Outputs are designed for clarity — structured insights and summaries that support engineering, product, and leadership decision-making.
• Built to Extend
Pipelines are designed to support future integrations (Slack, Notion, Sheets, and CRMs) without redesigning the core system.
• Built to Extend
Pipelines are designed to support future integrations (Slack, Notion, Sheets, and CRMs) without redesigning the core system.
Additional Projects:
Additional Projects:
These projects support and extend the same automation and orchestration patterns used in my flagship wwork, focusing on structured data flow, reliability, and real-world integration.
• PIE (PreMortem Integrity Engine) — Validation and audit layer for GenAI pipelines
Designing and implementing schema compliance checks and execution health monitoring to validate pipeline correctness and surface failures early.
• PIE (PreMortem Integrity Engine) — Validation and audit layer for GenAI pipelines
Designing and implementing schema compliance checks and execution health monitoring to validate pipeline correctness and surface failures early.
• PRS (PreMortem Runtime Supervisor) — Runtime enforcement and failure handling
Implementing retry policies, execution rules, and guardrails for multi-stage orchestration in production-like consditions.
• PRS (PreMortem Runtime Supervisor) — Runtime enforcement and failure handling
Implementing retry policies, execution rules, and guardrails for multi-stage orchestration in production-like consditions.
• BroadusBot — Multi-model task routing interface
Investigating governed task routing and role-based behavior across multiple LLMs.
• BroadusBot — Multi-model task routing interface
Investigating governed task routing and role-based behavior across multiple LLMs.
• Collabr8 — Structured intake and review workflows
Prototyped collaborative workflows for structured submission, review, and decision-making across teams.
• Collabr8 — Structured intake and review workflows
Prototyped collaborative workflows for structured submission, review, and decision-making across teams.
• Leadflow AI — Automated Client Intake & Scoring Pipeline
Prototyped webhook-driven ingestion and LLM-based scoring to evaluate structured automation patterns.
• Leadflow AI — Automated Client Intake & Scoring Pipeline
Prototyped webhook-driven ingestion and LLM-based scoring to evaluate structured automation patterns.
Engineering Skills
Engineering Skills
AI & Automation:
• LLM engineering (structured outputs, JSON schemas, multi-agent pipelines)
• LLM engineering (structured outputs, JSON schemas, multi-agent pipelines)
• Workflow orchestration and event-driven automation
• Workflow orchestration and event-driven automation
• Schema-validated scoring pipelines
• Schema-validated scoring pipelines
Backend & Integration:
• REST APIs, webhooks, schema validation
• REST APIs, webhooks, schema validation
• Data normalization and contract-driven interfaces
• Data normalization and contract-driven interfaces
• Secure configuration and environment management
• Secure configuration and environment management
Reliability & System Design:
• Modular, layered system architecture
• Modular, layered system architecture
• Fault tolerance, retries, and validation gates
• Production-ready documentation and reporting structures
• Fault tolerance, retries, and validation gates
• Production-ready documentation and reporting structures
I specialize in designing reliable, modular AI automation systems that integrate cleanly into real engineering workflows.
Tools & Technologies
Tools & Technologies
AI & LLM Development:
• OpenAI GPT-5.2, GPT-5.1, GPT-4.1, GPT-4o
• OpenAI GPT-5.2, GPT-5.1, GPT-4.1, GPT-4o
• Structured outputs (JSON schemas, function calling)
• Structured outputs (JSON schemas, function calling)
• Deterministic prompting with schema validation
• Deterministic prompting with schema validation
Automation & Orchestration Platforms:
• Pipedream (multi-step workflows, API integrations)
• Pipedream (multi-step workflows, API integrations)
• Make.com (process automation, scheduling , CRM workflows)
• Make.com (process automation, scheduling , CRM workflows)
• Event-driven automation (webhooks, triggers, conditional routing)
• Event-driven automation (webhooks, triggers, conditional routing)
Backend & Integration:
• REST APIs, webhooks, schema validation
• REST APIs, webhooks, schema validation
• Secure environment configuration and secrets handling
• Deterministic scoring and data normalization utilities
• Secure environment configuration and secrets handling
• Deterministic scoring and data normalization utilities
Data & Integrations:
• Google Sheets, Notion (structured data & workflows)
• Slack API, Twilio, email-based automations
• Google Sheets, Notion (structured data & workflows)
• Slack API, Twilio, email-based automations
Documentation & Design:
• Technical documentation and SOPs
• Architecture diagrams and workflow mapping
• Technical documentation and SOPs
• Architecture diagrams and workflow mapping
System Diagrams
System Diagrams
High-Level Architecture Overview (HLAO)
High-Level Architecture Overview (HLAO)
Core Pipeline Flow Diagram
(E2E Execution Flow)
Core Pipeline Flow Diagram
(E2E Execution Flow)
Engineering Case Studies
Engineering Case Studies
PREMORTEM AI — Automated Project Risk Intelligence System
AI-powered system that converts unstructured project inputs into standardized, repeatable risk analysis.
The Problem:
Project risk analysis is often manual, inconsistent, and subjective. Teams rely on free-form notes, ad-hoc scoring, and time-consuming documentation, which leads to missed risks and unclear decisions.
The Solution:
PreMortem AI automates the entire risk analysis workflow, producing a structured, end-to-end risk report from raw project descriptions.
Outputs include:
• Likelihood / impact scoring
• Risk clustering and theme detection
• Actionable mitigation strategies
• Executive-ready summaries
All generated through a deterministic, schema-validated pipeline.
My Role:
AI Systems Engineer
Designed the system architecture, LLM pipeline, and multi-agent workflow. Focused on structured outputs, validation, reliability, and clear documentation.
How it works:
A multi-stage LLM pipeline processes inputs through discovery, scoring, clustering, mitigation, and summarization stages.
Execution is orchestrated with schema validation, retries, and guardrails to ensure consistent, machine-usable outputs.
Results are written to:
• Google Sheets (operational data)
• Notion (long-term storage)
Tech:
GPT-5.2, GPT-5.1, GPT-4.1, Pydantic, JSON schemas, Pipedream, Notion API, Google Sheets API
Status:
Core architecture and system logic complete; currently in testing and validation.
Work History
Work History
Agilent Technologies — LC Column Loader & Operations Technician
Worked in regulated, SOP-driven production environments, supporting high-precision scientific workflows, documentation accuracy, and operational reliability.
Amazon — Problem Solver (Process Flow Support)
Diagnosed and resolved system-level workflow issues in high-volume operations, improving real-time execution and throughput.
B.O.S.S. 88 — Operations Support
Supported cross-team operations, troubleshooting execution gaps and improving process consistency and communication.
About Me
About Me
I’m an AI Software Engineer focused on building reliable, structured automation systems using LLMs and modern orchestration tools.
My background in regulated scientific operations and high-volume production environments shapes how I approach engineering, with an emphasis on consistency, validation, and predictable system behavior rather than demos or one-off prototypes.
I enjoy turning messy, unstructured inputs into production-ready pipelines, prioritizing schema validation, modular design, and end-to-end reliability. I learn quickly, iterate thoughtfully, and care about building systems that are easy to understand, maintain, and extend.
Get in Touch
Get in Touch
If you’re exploring automation systems, LLM integrations, or AI-driven workflow design feel free to reach out.
I'm always open to connecting.
Open to engineering discussions, technical collaboration, and automation architecture work.
Page updated
Google Sites
Report abuse