PRJ_003
003
Pro Forma — Full Stack
Eagle Venture Lab · AWS · Houston, Texas · Sep, 2025 – Jun, 2026
Overview

Pro Forma is a fully serverless pitch competition platform built for Eagle Venture Lab, Houston. As the full-stack engineer I designed and built the complete system: a React/Redux frontend, a REST API layer on AWS API Gateway, compute on Lambda, persistent storage in DynamoDB, and file assets on S3. The platform serves three user types — startup founders, judges and administrators — each with distinct data access patterns and UI flows, all communicating through a serverless API designed from scratch.

The Problem

Serverless architectures surface complexity in unexpected places — cold starts, IAM policies, DynamoDB access patterns, and Lambda function boundaries all require deliberate upfront design. On top of the technical architecture, requirements changed substantially and repeatedly throughout the engagement. Building on Lambda forced function-level clarity that actually helped manage this volatility: each domain (profiles, competitions, judging, rounds) lived in its own Lambda, making changes surgical rather than systemic.

Key Metrics
0
Servers managed (serverless)
AWS
Cloud platform
Lambda
Compute per domain
DynamoDB
Primary data store
Process & Timeline
Phase 1
Architecture & data modelling
Designed the serverless architecture and modelled DynamoDB access patterns up front to avoid costly redesigns once Lambda boundaries were set.
Phase 2
Lambda & API Gateway build
Built Lambda functions per domain behind API Gateway routing, defining clear function boundaries and least-privilege IAM policies.
Phase 3
Storage & integration
Wired S3 for company profile and pitch assets, connected the React frontend, and integrated the domain Lambdas into end-to-end flows.
Phase 4
Hardening & deployment
Addressed cold starts, tightened IAM policies, and deployed the full serverless stack for the live competition lifecycle.
Tech Stack
ReactReduxTailwind CSSSCSSJavaScriptAWS LambdaAPI GatewayDynamoDBAmazon S3Serverless Framework
Critical Self-Evaluation
This engagement was a graduate course in requirements management disguised as a software project. The serverless stack was the right choice — zero infrastructure overhead, predictable costs at low scale, and clean Lambda boundaries that absorbed a lot of the churn. What I would change architecturally: lean more heavily on DynamoDB single-table design from day one rather than retrofitting it. Access patterns evolved as the product evolved and backfilling index design is always more painful than it looks. On process: formalise a change request log from sprint one. Requirements volatility is normal in early-stage products — but naming it explicitly as a cost, with a paper trail, changes the conversation.