Skip to content
Sudhanshu Ratna Thakur

Sudhanshu Ratna Thakur

Delhi · Remote / Global

Lead II · UST

Production backend systems for regulated products. Applied AI with inspectable proof.

Hands-on technical lead with 12+ years delivering Spring Boot and WebFlux APIs, payment orchestration, and cloud services across banking and payments. Through UST, I support a Capital One UK programme after leading backend delivery for TSYS / Global Payments—and build source-backed applied-AI prototypes alongside production work.

Review selected workEmail meDownload CV
Proof linksGitHubHealth UI(sign-in)Materials UI(sign-in)LinkedIn

12+ years

Engineering since 2013

2 UK programmes

Delivered through UST

3 public AI repos

Inspectable source

AWS certified

Developer – Associate

Career evidence map

Regulated backend

Reporting APIs · payment routing

Delivery ownership

Design · test · release · support

Applied AI proof

3 public, inspectable repositories

2 regulated fintech programmes3 public AI repositories

Selected work

Production systems. Public AI proof.

Two regulated fintech programmes show production backend leadership. Public repositories show document retrieval, MCP-enabled search, and LLM-assisted risk review—with maturity and evidence stated on every card.

Role-secured reporting APIs
Production client deliveryCapital One UK · UST client programme

Secure Reporting API Modernization

Regulated reporting · production delivery

Backend architecture and product development for a SAS-to-Salesforce reporting migration, exposing governed Snowflake data through role-secured services across three operational reporting domains.

Business impact

CV-reported outcomes: enabled full SAS decommissioning, improved performance by approximately 35-40%, and reduced costs by approximately 30%.

See delivery contextDiscuss this system
View my role and technical evidence

Led Spring Boot and WebFlux API design for Fraud Security Tool, Section 78, and Exception Reports; implemented RBAC, event/workflow patterns, TDD/DDD, containerization, performance benchmarking, and Bedrock/PyPDF document-intelligence prototypes.

  • Backend architecture for SAS-to-Salesforce/Snowflake reporting
  • Secure RBAC plus event-driven and workflow-based services
JavaSpring BootWebFluxSpring SecurityEvent-driven architecture

Client-confidential work · original architecture view

CV-documented delivery through UST · quantified outcomes are CV-reported

Medical report retrieval
Medical Report Retrieval Assistant product and system view
Public source · hosted UI gatedPublic-source prototype

Medical Report Retrieval Assistant

Semantic retrieval · extractive QA

A Streamlit prototype for PDF ingestion, semantic retrieval, and extractive question answering over uploaded medical reports.

Business impact

Turns a static report into a searchable, question-led review flow so users can locate relevant document context without scanning every page.

Inspect retrieval sourceOpen hosted UI
View my role and technical evidence

Built MiniLM embeddings, FAISS retrieval, Hugging Face question answering, the upload-and-explore interface, and Cloudflare R2 storage with an S3 fallback.

  • PDF ingestion and semantic indexing
  • Question-led context retrieval
PythonStreamlitHugging FaceMiniLMFAISS

Verified public source · hosted Streamlit UI requires sign-in

Payments routing layer
Production client deliveryTSYS / Global Payments · UST client programme

TSYS Payments & Batch Platform Modernization

Payments · routing · batch automation

Backend leadership across a payment-routing platform and modernization of a legacy correspondence batch system serving global payment operations.

Business impact

CV-reported outcomes for the Spark/Airflow modernization: approximately 50-60% higher operational efficiency and approximately 70% less manual intervention.

See delivery contextDiscuss the architecture
View my role and technical evidence

Led payment routing, REST/SOAP/MQ/mainframe integration, Kafka services, Spark/Airflow batch modernization, secure APIs, encryption/decryption, validation frameworks, code quality, and delivery governance.

  • Spark and Airflow correspondence batch modernization
  • CV-reported: ~50-60% efficiency improvement and ~70% manual reduction
JavaSpring BootRESTSOAPMQ

Client-confidential work · original architecture view

CV-documented Java Tech Lead delivery through UST · quantified outcomes are CV-reported

MCP materials workflow
Public source · hosted UI gatedPublic-source prototype

Remote MCP Materials Research Platform

MCP tools · scientific search

A decoupled FastMCP/FastAPI and Streamlit workflow for materials lookup, property filtering, comparison, semantic search, and export.

Business impact

Brings search, lookup, comparison, and export into one workflow so researchers do not need to stitch those actions across separate tools.

Inspect MCP architectureOpen hosted UI
View my role and technical evidence

Designed reusable FastMCP tools, Materials API retrieval, MiniLM/Qdrant search, the remote service boundary, and the comparison/export experience.

  • Reusable MCP tools over scientific data
  • Separated Streamlit client and remote backend
PythonStreamlitFastAPIFastMCPMaterials API

Verified public source · hosted Streamlit UI requires sign-in

Manual change-risk review
Public prototypePublic manual-input prototype

LLM-Assisted Change-Risk Review

Change-risk classification · human review

A Streamlit prototype that turns pasted deployment or infrastructure-change context into a risk class, rationale, and mitigation guidance.

Business impact

Gives engineering teams a consistent pre-deployment discussion aid while keeping automated PR/log ingestion clearly separated as roadmap work.

Inspect prototype source
View my role and technical evidence

Implemented the working manual-input flow with Groq-hosted Llama analysis, structured output, feedback capture, and risk-trend views.

  • Manual change-context input
  • Structured risk rationale and mitigations
PythonStreamlitGroqLlamaChange-risk classification

Verified public source · working manual-input flow

Document intelligence MVP
Paperwake Document Intelligence MVP product and system view
Private working MVPPrivate working prototype

Paperwake Document Intelligence MVP

Document Q&A · extraction · human review

One Streamlit workspace for cited document Q&A, invoice extraction, contract review, synthetic samples, reviewer states, and exportable results.

Business impact

Packages three document-heavy workflows into inspectable product journeys that make source evidence, exceptions, and human review visible to stakeholders.

Request private walkthrough
View my role and technical evidence

Built OCR and PDF ingestion, MiniLM/NumPy retrieval with lexical fallback, provider-optional answers, structured extraction, citations, review states, and JSON/CSV export.

  • Cited document question answering
  • Invoice and contract extraction workflows
StreamlitPythonOCRPyPDFMiniLM

Local working Streamlit MVP · walkthrough available

See applied AI maturity and demos

Day-one ownership

What I can own on day one

12+ years of specific delivery evidence across regulated APIs, payment orchestration, technical leadership, and source-backed applied-AI prototypes.

Regulated API modernization

On UST's Capital One UK engagement, I designed and implemented role-secured Spring Boot and WebFlux APIs, integrated Snowflake data for Salesforce consumers, and built JMeter and Gatling performance tests.

Fraud Security Tool · Section 78 · Exception Reports

Payments orchestration leadership

As Java Tech Lead, I led backend delivery for TSYS / Global Payments across REST, SOAP, MQ and mainframe mappings, Kafka services, code reviews, LLD, and QA/UAT releases.

UST client engagement · Oct 2020 - Dec 2024

Applied AI with inspectable source

I built public repositories for health-document retrieval with MiniLM and FAISS, materials search with FastMCP and Qdrant, and a Groq/Llama change-risk prototype.

Three public repositories · hosted access labelled clearly

Architecture-to-operations ownership

I stay hands-on from requirements and LLD through implementation, reviews, TDD, performance testing, CI/CD, QA/UAT deployment, monitoring, production support, and mentoring.

Lead II · reviewer · mentor · release owner

12+ years

Engineering since 2013

Lead II

Current UST grade

4+ years

TSYS technical lead

AWS certified

Developer – Associate

Experience

Backend leadership across regulated products

Engineering since 2013, with current Lead II ownership on a Capital One UK programme and four years as Java Tech Lead for TSYS / Global Payments through UST.

  1. Jan 2025 - Present

    Current

    UST Global · Capital One UK engagement

    Lead II - Software Engineering

    Client role: Full Stack Developer

    In role
    Delhi, IndiaProgramme: Nottingham, UK · Remote, UK-aligned client delivery

    Leading backend architecture and product development within a UST engagement migrating Capital One UK's legacy SAS reporting to Salesforce-integrated services backed by Snowflake.

    Key contributions

    • Led backend architecture and product development for Fraud Security Tool, Section 78, and Exception Reports APIs
    • Implemented secure role-based access and Snowflake-to-Salesforce service integration
    • Applied event-driven and workflow-based architectures to improve modularity and responsiveness
    JavaSpring BootSpring WebFluxSpring SecurityREST APIsEvent-driven architectureWorkflow architectureSnowflake

  2. Oct 2020 - Dec 2024

    UST Global · TSYS / Global Payments engagement

    Java Tech Lead

    Delhi, IndiaProgramme: Leeds, UK · Remote, UK-aligned client delivery

    Led backend engineering for payment orchestration and a Spark/Airflow modernization of legacy correspondence batch processing within a UST client engagement.

    Key contributions

    • Modernized failure-prone correspondence batch processing with Apache Spark and automated Airflow workflows
    • CV-reported: improved operational efficiency by ~50-60% and reduced manual intervention by ~70%
    • Designed secure APIs, encryption/decryption services, and validation frameworks
    JavaSpring BootHibernateKafkaApache SparkApache AirflowEncryption / DecryptionValidation frameworks

  3. Mar 2018 - Sep 2020

    Publicis Sapient

    Senior Associate

    India

    Built backend services and Angular interfaces for a data-driven agritech platform supporting seed advisers and farm-planning workflows.

    Key contributions

    • Developed REST APIs and asynchronous data-processing flows for agribusiness tools
    • Implemented secured API access with Spring Security and OAuth2
    • Integrated Azure SendGrid for application email workflows
    Java 8Spring BootSpring SecurityOAuth2JPAHibernateREST APIsAngular

Earlier career

Enterprise foundations, 2013-2018

Full-stack delivery, workflow automation, reporting, and production support

  1. Oct 2016 - Feb 2018

    Consultant / Solutions Architect

    Capgemini · Johnson & Johnson engagement

    Delivered full-stack features for a life-sciences supply-chain and clinical-trial approval workflow in a Capgemini client engagement.

    Java 8 · Spring Boot · AngularJS · JPA

  2. Apr 2015 - Jun 2016

    Software Engineer

    R Systems

    Built alert-reporting automation that stored operational history in a relational database and presented it through interactive dashboards.

    Java 7 · Spring MVC · JPA · REST APIs

  3. Jul 2013 - Apr 2015

    Senior Programmer / Technical Lead

    CyberQ Consulting

    Built enterprise portal modules, dashboards, reporting queries, and audit tooling for a labour-inspection compliance platform.

    Java 7 · Spring · Hibernate · MySQL

Capability map

Evidence over keyword volume

A focused view of production delivery, applied projects, and supporting experience—without arbitrary proficiency percentages.

Production / lead deliveryApplied project experienceSupporting experience

Backend & API Delivery

Production / lead delivery

Primary production strength built across 12+ years of secure services, enterprise integration, and full-stack API delivery.

Capital One UK reporting APIs, Publicis agritech services, and Johnson & Johnson workflow delivery

View technologies
JavaSpring BootSpring WebFluxSpring SecurityREST APIsMicroservicesJPA / HibernateOAuth2Encryption / DecryptionValidation frameworksAngular 16TypeScript

Payments, Streaming & Data

Production / lead delivery

Distributed integration across payment channels, modern APIs, legacy systems, event streams, and enterprise data platforms.

TSYS / Global Payments orchestration plus Snowflake-to-Salesforce delivery for Capital One UK

View technologies
KafkaApache SparkApache AirflowEvent-driven architecturePayments orchestrationREST / SOAP / MQSnowflakeOraclePostgreSQLSQL reportingAsynchronous processingCQRS

Cloud, Delivery & Reliability

Production / lead delivery

Containerized delivery, CI/CD, automated testing, and operational tooling across production client programmes.

AWS-certified delivery using Kubernetes, Jenkins, Argo CD, TDD/DDD, performance testing, and production support

View technologies
AWSMicrosoft AzureGoogle CloudDockerKubernetesJenkinsArgo CDJUnit / MockitoCucumberJMeter / GatlingTDD / DDDCloudWatchSplunk

Applied AI Prototypes

Applied project experience

Document retrieval, tool-oriented search, and operational-analysis workflows demonstrated through public source and CV-documented applied work.

Verified Health, Materials, and SRE repositories plus Bedrock/Claude document automation experience

View technologies
PythonStreamlitMiniLMFAISSFastMCPQdrantPyPDFAmazon Bedrock / ClaudeLangChainRAG patternsboto3

Mentoring & community

Engineering judgment shared beyond delivery

Technical mentoring, volunteer architecture review, and continued learning across engineering and commercial communities.

Startup mentoring

Senior Technical Mentor

NASSCOM 10,000 Startups

Mentored early-stage SaaS and enterprise ventures on backend architecture, microservices, AWS scalability, and responsible GenAI foundations.

View focus
  • Backend architecture, microservices, AWS scalability, and engineering practices.
  • Responsible GenAI adoption grounded in secure, scalable backend foundations.
Open NASSCOM community
Platform review

Volunteer architecture reviewer

TiiQu C.I.C.

Reviewed a trust-centric platform's architecture, service dependencies, technical debt, and scalability risks.

View focus
  • Documented components, service interactions, and operational dependencies.
  • Identified technical-debt and scalability risks, with recommendations for onboarding, transparency, and planning.
Visit TiiQu
Community & commercial perspectiveEvents and public learning

Enterprise technology

DTX Manchester 2025

Knowledge-sharing across AI, cloud, digital transformation, and enterprise systems.

Commercial perspective

International Franchise Show London 2025

Attended at ExCeL London; the CV records 250+ exhibitors and 60+ seminars spanning franchising, entrepreneurship, and business expansion.

Open development

Public GitHub learning

Independent AI, backend, cloud-native, and developer-focused projects supporting continuous learning and public contribution.

Credentials & educationCloud, ML, compute, and engineering
  • AWS Certified Developer - Associate (2024-2027)
  • Machine Learning & Data Science Certification
  • Supercomputing & Computational Biology - IIT Delhi

Bachelor of Technology (B.Tech)

Gautam Buddha Technical University (GBTU) · 2007-2011

InterestsWhat keeps the learning loop active
GenAI & agentic workflowsTechnology communities & open sourceTravelTechnology & business readingFitness & outdoorsMusic & documentaries

Applied AI lab

Applied AI workflows, clearly staged

Source-backed prototypes, gated interfaces, and clearly labeled private or conceptual work across document, research, risk, and financial-review workflows.

Public source · UI gated

Medical Report Retrieval Assistant

PDF ingestion, MiniLM embeddings, FAISS retrieval, object-backed indexing, and extractive question answering in a Streamlit interface.

Business value

Turns a static report into a searchable, question-led review flow so users can locate relevant context without scanning every page.

Inspect retrieval sourceOpen hosted UI

Public source; hosted Streamlit UI currently requires sign-in

Public source · UI gated

Remote MCP Materials Research Platform

A tool-oriented scientific materials workflow for search, lookup, comparison, and export through a Streamlit interface.

Business value

Brings search, lookup, comparison, and export into one workflow so researchers do not need to stitch those actions across separate tools.

Inspect MCP architectureOpen hosted UI

Public source; hosted Streamlit UI currently requires sign-in

Public prototype

LLM-Assisted Change-Risk Review

Manual pasted-text deployment-risk analysis using a Streamlit interface and Groq-hosted Llama model.

Business value

Gives platform teams a consistent pre-deployment discussion aid while keeping automated PR and log ingestion clearly identified as roadmap work.

Inspect prototype source

Public source; automated PR and log ingestion remains roadmap work

Private working MVP

Paperwake Document Intelligence MVP

A private working Streamlit MVP combining cited document Q&A, invoice extraction, contract review, sample data, reviewer states, and export.

Business value

Packages three document-heavy workflows into inspectable journeys that make source evidence, exceptions, and human review visible.

Request walkthrough

Private working MVP; contact me for a guided walkthrough

Concept

Traceable Credit Review Copilot

A proposed financial-statement review workflow combining structured extraction, source citations, retrieval-assisted analysis, and human approval.

Business value

Frames AI as accountable decision support for document-heavy credit review, with traceability and oversight built in.

Discuss the concept

Concept; available for a discovery conversation

Hosted Streamlit interfaces require sign-in. Source is public where listed; private and concept walkthroughs are available by email.

Open GitHub profile

Open to senior opportunities

Hiring for backend, platform, or applied AI?

I'm available for senior roles where dependable backend systems, cloud delivery, and practical AI need to ship together.

Email SudhanshuDownload CVLinkedInGitHub
Senior Backend / Platform EngineerApplied AI EngineerJava / Cloud Technical Lead