Engineering Portfolio • Systems Architecture • Verified Impact

Engineering Portfolio & Systems Architecture

Over a decade of designing sub-second distributed software, eliminating millions in enterprise cloud waste, and shipping resilient production platforms across global organizations.

Welcome to my working portfolio. As a software builder and systems architect, I focus on two core engineering disciplines: building software that responds in under 350 milliseconds using fast computer memory, and removing wasteful server spend through disciplined FinOps governance. Below you will find the verified architecture projects, enterprise milestones, technical specifications, and production codebases I have led across Deloitte, Clean Harbors, PayPal (via eClerx), CAI, and Fraoula.

$12,000,000+
Cloud Waste Cut

Annual recurring infrastructure savings delivered at CAI.

10,000+ Hours
Dev Time Reclaimed

Automated testing pipelines and release engineering for PayPal payment systems.

99.5% Uptime
Production SLA

Sustained platform availability for Clean Harbors field logistics portal.

<350 ms
Audit Engine Latency

In-memory cryptographic compliance runtime at Fraoula with zero disk leaks.

Engineering Standards • How I Build

Architectural Philosophy & Core Operating Principles

High-leverage software engineering is not about using every new framework or adding complex microservice layers. It is about understanding fundamental hardware constraints, building clear boundaries, and making systems easy for human beings to read and maintain.

Principle 01 • Simplicity Over Novelty

Boring Technology Scaled Well

Boring, proven technologies like Linux, PostgreSQL, Redis, and plain HTTP scale reliably to millions of users when configured correctly. Novel tools introduce hidden failure modes, operational debt, and vendor lock-in that small teams cannot afford to debug during late-night outages.

Principle 02 • Zero Idle Compute

Every CPU Cycle Must Deliver Value

Unused server power is a sign of unmeasured code. When builders measure how their software runs and how much memory it actually needs, cloud costs drop naturally. Paying for servers that sit at 2 percent capacity is money thrown away.

Principle 03 • In-Memory Latency Budgets

RAM is Millions of Times Faster Than Disk

Accessing data in memory takes nanoseconds, while reading from disk or calling across network hops takes milliseconds. Designing software with fast in-memory indexes and streaming buffers eliminates 90 percent of latency bottlenecks before code ever reaches production.

Principle 04 • Observability Driven Decisions

No Architecture Without Telemetry

We never guess where a system is slow. Every architectural change starts with distributed tracing, latency histograms, and unit cost calculations. When you measure response times at the 99th percentile, you fix the real problems that customers actually feel.

Case Studies • Production Systems

Featured Enterprise Projects & Case Studies

Each project below represents a real production challenge solved using the Problem, Action, Result framework. These initiatives combine hands-on code development with high-level infrastructure strategy.

Enterprise FinOps • 2023-2024

CAI - $12M Cloud Rightsizing & Governance

Savings: $12M / Year

The Challenge: CAI operated an enormous multi-region AWS and Azure footprint serving critical enterprise software clients. Over several years of acquisitions and rapid engineering growth, the infrastructure accumulated thousands of unmanaged virtual machines, forgotten storage disks, and oversized database servers that inflated annual cloud invoices past acceptable limits.

The Architectural Action: I designed and implemented a company-wide FinOps governance framework. We introduced automated instance rightsizing scripts using Python and AWS CloudWatch metrics, enforced mandatory resource tagging policies via Terraform, converted predictable steady-state compute workloads to 3-year Savings Plans and Reserved Instances, and built an automated schedule engine that powered down non-production environments during nights and weekends.

The Business Result: The organization eliminated over $12,000,000 in annual recurring cloud spend while maintaining full operational stability. Engineering teams gained automated cost dashboards that provided real-time visibility into application unit economics.

Tech Stack: AWS (EC2, S3, RDS, CloudWatch), Azure Cost Management, Terraform, Python, Datadog
Read Full $12M Case Study →
FinTech • Release Automation • 2022-2023

PayPal (via eClerx) - Merchant Onboarding & CI/CD Pipeline

Efficiency: 10,000+ Hrs Saved

The Challenge: Global merchant integration suites suffered from long validation turnaround cycles. Engineers spent days manually verifying edge-case transaction payloads across different merchant currencies, causing deployment backlogs and delaying time-to-market for new payment capabilities.

The Architectural Action: I engineered automated end-to-end integration test harnesses and containerized test runners using Docker and Jenkins. We parallelized test executions across independent payment sandboxes, implemented intelligent test-caching layers, and created real-time webhook monitors to catch payload regressions before production staging.

The Business Result: Reclaimed over 10,000 hours of manual testing and engineering validation time per year. Release turnaround times were compressed by 40 percent, allowing developer teams to deploy updates daily instead of waiting for weekly freeze windows.

Tech Stack: Jenkins, Docker, Selenium, Python, REST APIs, Prometheus, Grafana
View Case Details →
Industrial Systems • High Availability • 2020-2022

Clean Harbors - Logistics Portal Modernization & 99.5% Uptime

SLA: 99.5% Uptime

The Challenge: Clean Harbors manages complex hazardous waste disposal and environmental response logistics across North America. Field dispatch teams relied on legacy portal infrastructure that suffered from recurring database lockups, unhandled network failures, and unpredictable downtime during severe weather operations.

The Architectural Action: Redesigned critical dispatch APIs using Java and Spring Boot, decoupled synchronized database calls using RabbitMQ message queues, and added Redis caching for regional truck routing schedules. Implemented automated health check circuits and multi-zone database read replicas to guarantee graceful failover.

The Business Result: Raised core platform availability to a strict 99.5 percent SLA, eliminating emergency field dispatch freezes and reducing dispatch synchronization latency by 65 percent across nationwide fleet operations.

Tech Stack: Java, Spring Boot, PostgreSQL, Redis, RabbitMQ, Docker, Kubernetes
Explore Architecture →
Document Pipeline • Performance Tuning • 2018-2020

Procentris - OCR Document Processing & Latency Reduction

Latency: 30% Faster

The Challenge: High-throughput document conversion pipelines processed tens of thousands of corporate legal and financial files daily. Under peak customer upload bursts, worker nodes encountered memory fragmentation, leading to frequent process terminations and long file backlog queues.

The Architectural Action: Refactored the core ingestion runtime from monolithic disk buffers to streaming memory chunks. Replaced unthrottled worker processes with bounded Redis Streams and worker concurrency throttles that preserved memory ceilings.

The Business Result: Cut end-to-end document conversion latency by 30 percent, eradicated process memory crashes, and enabled the platform to absorb 2.5x higher document volume without provisioning additional server instances.

Tech Stack: Node.js, Python, Redis Streams, Tesseract OCR, Linux System Tuning
View Case Study →
Enterprise Systems • Architecture Consolidation • 2016-2018

Deloitte Consulting - Cloud Transformation & Cost Reduction

Cost: 20% Reduction

The Challenge: Fortune 500 enterprise software clients faced escalating maintenance fees and fragile microservice boundaries following rapid legacy cloud migrations. Multiple duplicate services performed similar authorization and routing checks, introducing redundant latency hops.

The Architectural Action: Conducted comprehensive architectural audits of service dependencies, consolidated redundant microservices into cohesive core components, and established standardized API gateways with centralized token verification.

The Business Result: Slashed ongoing infrastructure operational expenses by 20 percent and reduced cross-service network hop latency by half across core user journeys.

Tech Stack: AWS ECS, Docker, OpenAPI, Microservices Architecture, Splunk
Review Details →
Active Venture • In Production • 2024-Present

Fraoula - In-Memory Cryptographic Compliance Engine

Latency: <real-time

The Challenge: Modern privacy regulations (GDPR, HIPAA, and CCPA) require businesses to inspect and verify outbound customer marketing records and payment receipts without exposing sensitive consumer identities to third-party databases or third-party log servers.

The Architectural Action: As Co-Founder and Chief Product Officer, I designed an in-memory inspection engine that verifies cryptographic hash proofs in fast RAM. The software checks data formatting and compliance rules in milliseconds and immediately wipes all working buffers from memory, guaranteeing zero disk leakage.

The Business Result: The engine executes full compliance inspections in under 350 milliseconds (faster than the blink of an eye), running reliably in live production across more than 20 partner environments.

Tech Stack: Rust, Node.js, Redis, In-Memory Cryptography, Cloudflare Workers
Visit Live Engine →
Core Capabilities • Technical Breakdown

Technical Domains & Competencies Matrix

A detailed overview of the technical domains, architectural patterns, and production toolchains I bring to engineering leadership and consulting engagements.

1. Cloud Cost Optimization & FinOps

  • Tag-based multi-account infrastructure cost allocation
  • Compute rightsizing and automated instance scaling
  • Reserved Instances and Savings Plans mathematical modeling
  • S3 and Azure blob storage lifecycle tiering
  • Database connection pooling and query optimization
  • Elimination of unattached volumes and idle NAT gateways

2. Low-Latency Distributed Systems

  • In-memory data structures, Redis, and RAM caching design
  • Asynchronous message brokers (RabbitMQ, Redis Streams)
  • Zero-copy streaming I/O and bounded queue management
  • REST and event-driven microservice orchestration
  • Cryptographic data validation without persistent disk storage
  • Sub-second API latency budget enforcement

3. Site Reliability & Observability

  • Strict SLA, SLO, and SLI definition and monitoring
  • High availability multi-region active-passive failover
  • Prometheus, Grafana, and Datadog telemetry dashboards
  • Automated disaster recovery drills and health probes
  • Distributed tracing and p95/p99 latency analysis
  • Graceful degradation and automatic fail-safe protections

4. Engineering Leadership & Delivery

  • Small-team leverage: shipping outsized results with 5 to 10 engineers
  • Plain-English technical specifications and architectural decision records
  • Legacy codebase stabilization and incremental refactoring
  • CI/CD automation and test parallelization for fast deploys
  • Mentorship for senior engineers and engineering managers
  • Bridge between technical teams and executive business leadership
Work Navigation • Topic Clusters

Explore Specific Projects, Products & Archives

Follow these direct links to dive deeper into my commercial products, technical research, transparent metrics, and published guides.

01 • Case Studies Hub

All 20 Production Case Studies

Comprehensive archive of all 20 real-world system optimization case studies.

View All Case Studies →
02 • Commercial Companies

Companies Built

Venture history including Fraoula and Sovereign Software Studio.

Explore Companies →
03 • Software Products

Product Gallery

Tour of software systems built under my product and architecture direction.

View Products →
04 • Transparent Metrics

Open Operational Numbers

Publicly shared operational statistics, business revenue, and readership metrics.

See Open Numbers →
05 • Engineering Failures

My Real Mistakes

Six major engineering and business failures and what they taught me about building software.

Read Mistakes →
06 • Technical Research

Technical Research

White papers on in-memory computer performance and distributed systems design.

Read Research →
Work With Arkajit • Advisory & Architecture

Need Help Optimizing Your Software or Cutting Cloud Bills?

I advise technology founders, CTOs, and engineering leaders on software architecture, eliminating cloud waste, and improving system response times. You can book a direct 45-minute advisory call or commission a comprehensive Speed & Cost Audit.

Advisory Session Fee: INR 2,000 / $20 USD
Book Advisory Session → Speed & Cost Audit ($8,000) Review Engagement Options