KG
0%

Portfolio · 2026

Bangalore, India · --:--:-- IST

Krishna Ganga

AI engineer building and operating LLM-powered systems in production. Agents, RAG pipelines, and the backend behind them.

Technical Consultant at EY · Published at ICSTE 2024

Selected work Get in touch

Currently working on

01 / Selected work

Things I’ve
shipped.

Product engineering, applied AI, and interfaces that make sophisticated things feel simple.

Pick one.

02 / Enterprise

Guidewire at EY.

PolicyCenter Developer

Technical Consultant · Ernst & Young · Bangalore

Apr 2025 — Present

I build and maintain production PolicyCenter for one of the largest property and casualty carriers in the United States, under the review, testing and release discipline of a regulated system.

Converted from a substitute to a full-time role on the strength of delivery within the Guidewire practice.

Recognition Achiever Extraordinaire Award Ernst & Young

  • Guidewire Associate Certified
  • GitHub Copilot Certified

  1. Configuration

    PCF screen modifications, data model extensions and product model configuration, shaped to the carrier's own underwriting rules.

  2. Gosu development

    New business logic across the Homeowners and Personal Auto lines, out of the box where the platform allows and extended where it does not.

  3. Policy lifecycle

    Quoting, binding, endorsements and renewals, taken from business and QA requirements through to code that holds up in review.

  4. Internal tooling

    ROLLOUT agents that validate their own generated configuration rather than handing it over unchecked, cutting roughly three days of manual work per rollout. Every team on the project uses them.

03 / Capabilities

The toolkit.

Languages
PythonJavaGosuSQLSwiftJavaScript
LLMs & agents
Prompt engineeringContext managementTool use & function callingRAG pipelinesStructured output parsingSchema validationGuardrails & refusal behaviourFailure-mode debuggingClaude APIGemini APIOpenAI API
Evaluation & observability
Latency benchmarking (median, p90)Extraction accuracy scoringAdversarial guardrail suitesManual review scoringVersioned normalizationRegression attributionRanked failure capture
Backend & infrastructure
Cloudflare WorkersR2 object storageSupabase / PostgreSQLSchema designRow-level securityREST integrationWranglerServerless deploymentCI/CD
ML & data
Deep learningCNNsComputer visionTensorFlowKerasOpenCVMediaPipeNumPyPandas
Application & tooling
SwiftUICombineStreamlitGit & GitHubClaude CodeGitHub Copilot certifiedVS CodeJupyter

04 / About

I build and operate LLM-powered systems in production: agents, RAG pipelines, tool integrations, and the backend services behind them.

At EY I deliver Guidewire platform work for enterprise insurance clients, where internal agents I built cut roughly three days of manual configuration per rollout. Outside it I ship my own products end to end, design through backend.

0Products built