Skip to content
Chandrakanth Thadkapally
Systems built

Different industries, one recurring shape.

A commerce catalogue, a staffing platform, a payroll run and a settlement file are the same problem wearing different clothes: two systems hold different beliefs about one event, and something has to decide which is right, exactly and explainably. Having built all four is what makes the pattern visible — and it is why the agent work is the natural next step rather than a change of subject.

Described by scope and scale only. No internal system names, vendors, contract terms, volumes, thresholds or business rules appear here — not paraphrased, not anonymised.

§01
Where I go deepest
Two domains carry most of the work, and they are the two the current role sits across.
  1. 01
    2020 — present
    Current

    Retail & e-commerce

    A catalogue nobody can fully search, a platform too coupled to change safely, and a release cadence that decides how fast the business can move.

    Five years on a commerce platform and now at retail scale. The recurring work is decomposition — taking a system that was one thing and making it several, without losing the guarantees the business had come to rely on — and then building the delivery practice that keeps the pieces shippable.

    Built
    • Decomposition of a monolithic commerce platform into a service-oriented architecture, with the migration sequenced so the business kept trading throughout.
    • Continuous delivery pipelines for Kubernetes-based infrastructure, taking deployment from an event to a routine.
    • Centralised configuration and secret management across environments, removing the drift that makes staging a poor predictor of production.
    • Retail correction systems — the machinery that fixes what the happy path got wrong.
    • Search architecture: the four-stage retrieval, rewriting, ranking and reranking pipeline and its latency budget.
    Where

    Walmart Global Tech · Nebraska Furniture Mart

    Stack

    Java · Spring Boot · .NET Core · Kafka · Azure · Kubernetes · React

  2. 02
    2025 — present
    Current

    Payments & financial operations

    Two systems disagree about the same economic event, and the disagreement has to be resolved exactly, explainably, and in a way an auditor will accept.

    Payment reconciliation at a scale where the exception queue is a staffing decision rather than a bug list. The engineering is event-driven services on one side and data pipelines on the other, joined by an analytics surface that lets finance see that transactions are accurate instead of being told they are.

    Built
    • Event-driven services for large-scale payment reconciliation, built on Kafka with Java and Spring Boot.
    • Financial data pipelines spanning Google BigQuery and SQL Server.
    • Backend services in .NET along the reconciliation path.
    • Analytics through Power BI so transaction accuracy and compliance are observable rather than asserted.
    Where

    Walmart Global Tech

    Stack

    Kafka · Java · Spring Boot · BigQuery · SQL Server · .NET · Power BI

§02
Earlier work
Real experience, not the current focus — but it is where the pattern above came from.

  1. Healthcare staffing

    2018 — 2020

    AMN Healthcare

    Vendor management workflows rebuilt on .NET Core with a React and Redux front end, and access management consolidated onto OAuth 2.0 and OpenID Connect.

  2. HCM & payroll

    2018

    Ultimate Software

    Microservices inside a multi-tenant human capital management product spanning HR, compensation, talent, payroll and time-and-labour management.

  3. Telecom provisioning

    2017 — 2018

    Level 3 Communications

    A customer-facing product wizard matching broadband plans and quotes to stated preferences, on Angular and .NET Core Web API.

§03
Where this is written up

The essays

Nine long-form pieces on agents, evaluation, privacy, payments and commerce — each with an original diagram and a worked example on synthetic data.

Read the writing →

The research programme

Six strands with the status each has actually earned, and two complete manuscripts on graph-based and constraint-aware reconciliation.

See the programme →

The record

The factual employment history behind all of this, currently Senior Software Engineer at Walmart Global Tech.

Read the CV →

Building something in any of these areas? Get in touch. I read everything.