Tsvetan
Konov
I turn wild AI ideas into working prototypes — fast, and built to survive production.
$
probably debugging an agent
days
from “what if?” to working demo
70%
of paperwork handed off to agents
∞
evals — vibes aren't a test plan
0
slide decks where a demo would do
About
quality-obsessed, prototype-brained
By day I'm an AI Prototyping Engineer at Tokio Marine Global Services — turning “what if we used AI for this?” into working demos the business can actually poke at.
Before that, I spent 5+ years leading QA for enterprise fintech — the kind of work where one bad defect reaches 110,000 users. So my prototypes come with a rare feature: they don't fall apart when you touch them.
Agents, RAG pipelines, eval harnesses, LLM tooling — built fast, tested like they matter, demoed early.
[ current_status ]
- role:
- AI Prototyping Engineer
- company:
- Tokio Marine Global Services
- focus:
- agents · RAG · evals · LLM tools
- previously:
- Lead QA Engineer @ TCS
- education:
- B.S. IT, Rutgers — magna cum laude
- status:
- open to AI conversations
Work
where the receipts come from
AI Prototyping Engineer
2024 → NOWTokio Marine Global Services
- →Prototype LLM-powered tools and agentic workflows for insurance operations — from whiteboard sketch to working demo in days, not quarters.
- →Build RAG pipelines over policy, claims, and underwriting documents so teams can query institutional knowledge in plain English.
- →Bring QA-grade rigor to fast-moving AI experiments: evals, guardrail testing, and honest demos of what works and what doesn't.
- →Partner with business stakeholders to scope, demo, and iterate on the AI use cases actually worth scaling.
Lead Software QA Engineer
2021 → 2024Tata Consultancy Services
- →Owned QA strategy for a Fortune 100 client's mainframe-to-cloud transformation — defect-free delivery of cloud infrastructure and a modern dashboard.
- →Directed an end-to-end Selenium + Python automation framework that cut regression cycle time by 57%.
- →Ran defect triage, release sign-offs, and UAT across Agile sprints impacting 110K end users.
SDET — Software Engineer in Test
2019 → 2021Tata Consultancy Services
- →Automated end-to-end and regression testing with Java/Selenium.
- →Wired hundreds of test cases into Jenkins + GitHub CI/CD pipelines for Salesforce deployments.
- →Built automated data-provisioning scripts for repeatable, scalable test environments.
Technical Coordinator
2018 → 2019MicroTek
- →Automated environment setup with PowerShell and Bash, slashing provisioning times.
- →Resolved 90% of instructor technical issues within 10 minutes — a 25% lift in session completion rates.
Projects
click anything — it opens
★ featuredTickerSense
AI-powered stock trading platform
An intelligent trading platform combining machine learning with real-time market data. Python + React with live market feeds, sentiment analysis, automated trading signals, portfolio tracking, risk assessment, and AI-driven trade recommendations.
★ featuredIntelligent Document Processor
LangChain + RAG document automation
A document processing system built on LangChain and RAG architecture that extracts, analyzes, and categorizes business documents automatically. OCR, natural language understanding, and automated workflow triggers — cutting manual processing time by 70% for enterprise clients.
Mainframe-to-Cloud Transformation
Fortune 100 financial services
Test Automation Framework
57% faster testing cycles
QA Analytics Dashboard
Real-time test metrics & insights
Salesforce CI/CD Pipeline
Automated testing & deployment
AI Engineering Certification
Gen AI agents & LLM development
Toolkit
the stuff that gets used
[ ai / ml ]
[ qa / testing ]
[ languages ]
[ cloud / platform ]
[ delivery ]
[ human ]
Credentials
paper that backs it up
[ education ]
B.S. Information Technology
Rutgers University
Minor in Organizational Psychology
AI Engineer
AI Makerspace
AI Agent Builder
MindStudio
Salesforce Certified Administrator
Salesforce
Azure Cloud Fundamentals
Microsoft
Let's build
something.
Building with AI? Need a prototype by Friday? Met me somewhere and want to continue the conversation?
tsvetan.konov@gmail.com
in/tsvetan-konov
github
T-K-O-H
[ send_message ]