Illustration: a request moves through the interface, service and data layers to a response, and each layer reports to an observed trace. It doesn't show any employer's architecture.

Jaden Stittsworth · Senior Software Engineer · Pittsburgh, PA

I turn complex business workflows into production software.

I've owned distribution workflows and a care-agent application across the interface, backend services, and AWS.

Email me

Decisions behind the work.

Each case below runs from the problem to the result I verified, with my decisions in between.

I describe employer work at a high level, and their code and internal systems stay private. The workflow illustrations are reconstructions.

IRALOGIX · Product ownership · React · Go · AWS Lambda

Building the care-agent experience in Amazon Connect

Problem

On a call, agents needed account context, previous interactions, and the next action in front of them. Lookups kept timing out because of cross-region access and a cold cache.

Caller lookup→Account context→Agent workspace

Reconstructed workflow. Each night, the contact-center region's lookup cache refreshed from the first region's cache.

Illustration. On the left is the first region, where PostgreSQL stays the source of truth. On the right, the contact-center region's recreated lookup cache refreshes nightly from the first region's cache (dashed) and serves caller lookup into the agent workspace.

Ownership

I owned the agent-facing application and its supporting call workflows for the Salesforce-to-Amazon Connect migration. That covered the React interface, Go services, and AWS Lambda functions.

The account context had to reach the contact-center application in another AWS region before the caller got to an agent.

My decisions

With stakeholders, I organized the call view around that sequence. I recreated the Valkey lookup cache in the contact-center region and refreshed it nightly from the first region's cache.

PostgreSQL stayed the source of truth. If a refresh failed, the last good cache kept serving until a later refresh worked.

Verified result

4×

In recorded comparisons of the same scenarios, the backend account-lookup request got roughly 4× faster. That resolved the recurring timeouts.

IRALOGIX · Domain collaboration · Financial workflows · Full stack

Turning retirement-account rules into a usable distribution wizard

I owned the distribution wizard's workflow, interface, and backend behavior. With domain experts, I worked out which withdrawal paths were allowed and how withholding choices changed the request. I also decided which steps a customer had to revisit after changing an earlier answer.

Problem

Earlier answers changed later steps. The final request still had to match the customer's current selections.

Production follow-through

Some distributions sat at 'awaiting signature' after customers had finished DocuSign signing. I fixed them, reconciled the affected requests, and checked the whole signing-to-processing workflow.

Read the story

The workflow tied withdrawal choices to bank information, required authorization, and downstream processing. On the wider platform, I also worked on Plaid bank verification, DocuSign signing, and ACH processing.

When a customer changed a choice, I made them revisit the affected steps and fix anything missing or invalid before they could submit.

Fixing a signing-to-processing failure

In one production issue, a distribution still showed 'awaiting signature' after the customer had completed DocuSign signing.

I traced the mismatch across the portal, backend, and signing integration. Completion events weren't tied to the right originating request, so I corrected that association and repaired the status handling. Then I reconciled the affected requests and checked the whole signing-to-processing workflow.

My decisions

I turned the rules into branching, explanations, withholding calculations, validation, and backend behavior. Change an answer and you have to revisit the steps it affects.

Change an earlier answer

Scroll to fill in the steps. Then change a finished one and see which later steps need another look.

  1. Withdrawal path
  2. Withholding
  3. Bank information
  4. Required authorization
  5. Validated request

Changed answer→Revisit affected steps→Validated request

Reconstructed workflow.

Tata Consultancy Services / PNC Bank · Java Swing · Spring · Service integration

One application for test data across services owned by different teams

Problem

To get test data, employees had to find the right service, decipher its mnemonics, and supply its required parameters.

User choice→Service-specific requests→Multi-step calls→Created or repaired record

Reconstructed workflow. Service names and internal systems stay private.

50+QA engineerscreated, repaired, and customized test data through one interface

My decisions

I translated user inputs into service-specific requests, coordinated multi-step calls, and handled service errors. When a service lacked a capability we needed, I worked with its owners to resolve it.

Verified result

50+ QA engineers could create and modify test data through one application. Before, they had to decipher and call each service separately.

IRALOGIX · Java services · API design · Data integrity

Preserving saved data during account edits

Problem

The old update path treated an incomplete request as a replacement, so any omitted field erased its stored value.

Saved record→Apply supplied changes→Preserve omitted fields

Reconstructed update behavior. Omitted fields keep their existing values.

I eliminated a data-loss defect across Java services and two portals. I designed and implemented a partial-update API, then updated its clients so they preserved omitted fields.

I implemented the Java backend and the supporting frontend changes, and coordinated the shared library across the Staff and Partner portals.

The fix needed one consistent update contract across the API and its consumers.

IRALOGIX · AWS · Observability · Release verification

Making release readiness verifiable

The team made release decisions together. My part was building confidence in the changes I worked on and raising unresolved concerns before release.

Deployment checks→Production evidence→Shared release decision

Reconstructed verification process. Release decisions draw on deployment health and observed production behavior.

For scale: the IRALOGIX platform holds $1.3B under administration across 365,000 accounts. Those are company figures, separate from my individual impact.

I defined and ran the ECS-to-EKS cutover checks in Argo CD and Datadog, covering deployment health and production request behavior. I also wrote the team's verification playbook and took the findings into shared release decisions.

The playbook paired Argo CD deployment checks with Datadog evidence. It showed production requests reaching the new workloads, completing successfully, and staying within acceptable error and latency behavior.

31Production deployments I executed in 2025Zero critical incidents from those releases.

13Critical production bugs I resolvedEight were P0/P1 issues. Average turnaround was under 24 hours.

Shared tools & code review

Improving how teams work.

A shared interface operations wanted on more screens

Filtering, sorting, and pagination helped staff find the records that needed attention. With the relevant fields in the table, they could compare records before opening one. And since the screens browsed the same way, staff didn't have to relearn each one.

Reviews that looked at the whole workflow

Because I worked across portals and services, I could spot dependencies that are easy to miss when you review a change in isolation.

I built the shared Angular table that became the Staff Portal standard. Operations then asked for it on more screens.

Other developers could reuse it through the shared component library.

96Pull requests reviewedAcross 12 repositories for a 15-engineer team.

In review I caught API compatibility issues and required tests for realistic scenarios. I also steered developers toward shared implementations, which kept application behavior consistent.

AI-assisted delivery

Problem

Agent-written changes could pass simple tests and still fail in the application.

Making agent-written changes verifiable.

I built setup code and tailored test environments so agents could run the applications and supporting services, then exercised realistic workflows before accepting their changes. I retained responsibility for technical decisions, integration, and release readiness.

Experience & toolkit

Backend depth. A broader perspective.

I've worked in financial services and assessment software. I build applications, and the tools teams use to ship them.

  1. IRALOGIX

    October 2024 - August 2026

    Senior Software Engineer

    I was the primary engineer across four production systems on a retirement-account platform. I shaped workflows with Product, operations, and domain experts, then saw them through production release. In 2025 I executed 31 production deployments, with zero critical incidents from those releases.

  2. Talogy Tech / PSI Online

    2022 - 2024

    Software Engineer, SDET then Full-Stack Developer

    I built assessment workflows in Angular/TypeScript and Spring Boot, covering step progression, validation, and error handling. With the team, I found important workflows the tests missed and added automated tests as I built features.

  3. Tata Consultancy Services / PNC Bank

    2021 - 2022

    Software Developer / Lead Test Analyst

    I designed and built a Java Swing application backed by Spring. 50+ QA engineers used it to create, repair, and customize test data through one interface. I also connected services owned by different teams, worked with their owners on missing capabilities, and coordinated release testing.

Build
Java · Spring Boot · Go · TypeScript · React · Angular · Node.js · GraphQL · SQL
Connect & operate
AWS Lambda · Amazon Connect · RabbitMQ · PostgreSQL · Valkey · Docker · Kubernetes · Argo CD · Datadog
Verify
JUnit · Selenium · Cucumber · Integration & release testing
AI tooling
Python · Claude Code · MCP · Anthropic SDK · LLM evaluations

Tell me what your team is building and what's getting in the way.

Let's talk about your engineering team.

I'm looking for senior software engineering roles in backend, full-stack, platform, or AI engineering.

Email me