Libin Samkutty
Software Development Engineer in Test
4+ years ML exp
Python
SQL
JavaScript
TypeScript
Node JS
Accessibility & Semantics: 8/10
State Management & Data Flow: 8/10
UI Component Architecture: 6/10
Active 13 days ago
+91 (7802) 865503 Invite to interview
Message
Download CVCV
Overview
Technical skills
Timeline
Roles
Overview
Software Development Engineer in Test/AI Quality Engineer with 4.5+ years of experience specializing in Large Language Models (LLMs), Retrieval-Augmented Generation (RAG), and conversational AI platforms. Experienced in designing scalable evaluation frameworks, validating AI systems across multilingual and multi-turn interactions, and implementing production monitoring to improve the reliability, accuracy, and safety of AI applications. Strong expertise in LLM evaluation, hallucination detection, RAG pipeline validation, API automation, data validation, SQL, cloud technologies, and CI/CD, with a focus on combining software quality engineering and AI evaluation to deliver robust, observable, and production-ready AI systems.
Technical skills
Python
SQL
JavaScript
TypeScript• Senior
Node JS• Middle
Node JS
Dotenv
Databases
PostgreSQL
Apache Kafka• 3y+
Google BigQuery• 3y+
Frontend
React.js
Vite
Tailwind CSS
Zod
DevOps
Amazon EC2
AWS
CI/CD
CircleCI
Docker
GCP
Jenkins
Rest API
CI/CD
AWS Lambda• 3y+
GitHub Actions• 3y+
Grafana• 3y+
Prometheus• 3y+
QA
Pact
Playwright
Allure Reporting
Vitest
Selenium
QA
Postman• 4y+
Pytest• 4y+
Robot Framework• 4y+
AI/ML
Claude
Gemini
Hallucination
Vertex AI
LangChain
NER• 4y+
NLP• 4y+
Anomaly Detection• 3y+
DeepEval• 3y+
LangSmith• 3y+
LLM• 3y+
RAG• 3y+
Cybersecurity
OWASP ZAP
Timeline
Senior QA Automation Engineer
•
Senior
Jeavio India Pvt. Ltd.
•
Full-Time
Owned end-to-end AI quality strategy across multiple conversational AI product lines, including evaluation framework lifecycle decisions and direct client coordination. Advanced a deterministic PASS/FAIL judging approach with per-dimension criteria to replace CI thresholding, focusing on calibrated, reproducible evaluation gates. Maintained nightly evaluation CI pipelines and tracked evaluation scores during the transition to strict pass/fail gating.
GitHub Actions
DeepEval
LLM
RAG
QA Engineer
•
Middle
Jeavio India Pvt. Ltd.
•
Full-Time
Co-designed a streaming AI evaluation pipeline using Kafka-based ingestion, Lambda evaluators, LangSmith annotation/trace workflows, and Prometheus/Grafana monitoring. Implemented sampling and aggregation logic to detect clinical-safety regressions that were not caught by offline layers, improving the safety pass rate shortly after go-live. Built a multi-phase evaluation framework for correctness, reproducibility, and enforceable pass/fail gates, including LLM-as-a-judge calibration, multilingual golden datasets, and DeepEval G-Eval dimensions with BigQuery-backed multi-turn evaluation.
GitHub Actionssince 2023
AWS Lambda
Apache Kafka
LangSmith
Prometheus
Grafana
Anomaly Detection
Google BigQuery
DeepEvalsince 2023
LLMsince 2023
RAGsince 2023
Associate QA Engineer
•
Middle
Jeavio India Pvt. Ltd.
•
Full-Time
Established initial structured QA practices including test templates, coverage planning, defect governance, and severity classification to reduce defect escapes. Created Robot Framework UI automation and Pytest API automation suites, and improved CI smoke-run performance by addressing test sharding and scheduling. Expanded conversational AI and NLP intent classification testing, identifying high-severity defects and data anomalies before release.
Robot Framework
Pytest
NLP
Gujarat Technological University
Bachelor's Degree •
Computer Engineering
Project Intern
•
Junior
Jeavio India Pvt. Ltd.
•
Full-Time
Supported manual functional and integration testing for an NLP chatbot, including intent classification, NER flows, and API validation scenarios. Documented conversational routing behaviors and edge cases to aid testing and release readiness. Contributed to validation work using Postman during integration testing cycles.
Postman
NLPsince 2022
NER
Senior Frontend Developer
Confidence: High UI Engineer
Frontend UI engineer (senior-level) focused on building accessible, well-tested React applications and robust client-side state machines. The strongest proven skill is stateful front-end engineering and safe persistence, evidenced by wordwright/src/engine/reducer.ts and wordwright/src/storage/safeStorage.ts which show careful reducer design, runtime validation, and storage fallbacks. What is not evidenced is large-scale multi-team frontend systems leadership, SSR/edge backend work, or explicit bundle-level performance measurement artifacts.
UI Component Architecture
6/10
How interface parts are built
Clear, deliberate component and state boundaries with custom hooks and a reducer-driven game engine; evidence of purposeful architecture rather than copied UI-kit components.
Evidence
wordwright/src/engine/reducer.ts: gameReducer with well-scoped actions and purity tests
wordwright/src/hooks/useRevealTimeline.ts: custom hook coordinating reveal lifecycle
wordwright/src/state/contexts.ts: scoped React contexts for game, settings, stats
Responsive & Cross-browser
6/10
Works on all screens and browsers
Responsive design and cross-browser concerns are addressed with media queries and targeted e2e tests for OS/browser features and reduced-motion.
Evidence
wordwright/src/styles/index.css: responsive media queries and layout rules
wordwright/e2e/responsive.spec.ts: automated checks for responsive behavior
wordwright/e2e/edge-cases.spec.ts: tests that exercise OS-level preferences and offline behavior
Performance Optimization
4/10
Speed of the interface
Some performance attention (offline-first tests, Lighthouse claims in docs, careful storage handling) but limited explicit measured optimization artifacts like bundle analysis or before/after metrics in code.
Evidence
wordwright/src/engine/random.ts: seededRandom and deterministic selection to avoid runtime nondeterminism
wordwright/e2e/edge-cases.spec.ts: offline and request-capture tests validating no network calls after initial load
Accessibility & Semantics
8/10
Usable for everyone
Strong accessibility focus with axe-core audits in CI, keyboard and focus management tests, and ARIA/role-oriented e2e assertions demonstrating real a11y work on custom widgets.
Evidence
wordwright/e2e/a11y.spec.ts: Axe audits and detailed keyboard/focus tests
wordwright/src/hooks/useFocusTrap.ts: custom focus-trap hook used by dialogs and modals
State Management & Data Flow
8/10
Managing data in the app
Robust server-independent state discipline and storage safety: typed runtime validators, safe storage fallbacks, idempotent stats aggregation and clear handling of corrupt or unavailable LocalStorage.
Evidence
wordwright/src/engine/reducer.ts: thorough reducer logic with purity and edge-case tests
wordwright/src/storage/safeStorage.ts: probe/createSafeStorage with backend resolution and fallbacks
wordwright/src/storage/schemas.ts: runtime validators for settings, stats, sessions and migration resilience
UX & Visual Polish
6/10
Look and feel quality
Thoughtful UX details - storage warnings, offline behavior, reduced-motion handling and automated visual screenshots - showing attention to perceived performance and resilient UX flows.
Evidence
wordwright/e2e/screenshots.spec.ts: visual screenshot tests exercising UI states
wordwright/src/styles/index.css: careful styling and transitions, plus prefers-reduced-motion support via tests
Expertise
React• Senior
Frontend Architecture & Build Tools• Senior
Industries
Artificial Intelligence• Senior
Technologies
Tailwind CSS
React.js
Vite
Zod
Recommendations
- Lead development of accessible single-page React applications that need robust client-side state machines and offline resilience.
- Implement complex UI components and custom hooks where keyboard/focus management and a11y matter, such as dialog systems, game UIs or accessible widgets.
- Build and own front-end test harnesses and e2e pipelines (Playwright) that validate UX, accessibility and cross-browser edge cases.
- Drive component- and state-driven refactors that centralize storage validation and migration logic for web apps that must survive upgrades and corrupt storage.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
Senior QA Engineer
Confidence: High SDET
Senior SDET and test-engineering practitioner specializing in robust web and API automation frameworks. The strongest proven skill is building production-grade Playwright and API test infrastructure and cross-layer tests, as shown by playwright-web-api-automation-ts/playwright.config.ts and tests/api/notes-crud.spec.ts. There is limited public evidence of large-scale distributed performance engineering, formal incident reports, or external contract-test orchestration in the analyzed human-authored files.
Test Automation Frameworks
8/10
Building automated tests
Strong, production-ready test automation framework design: Playwright projects with page objects, custom fixtures, parallelization and Allure reporting show deliberate infra and selector discipline.
Evidence
playwright-web-api-automation-ts/playwright.config.ts: central Playwright config with reporters, retries, workers and snapshotDir
playwright-web-api-automation-ts/src/fixtures/page-manager.ts: custom PageManager fixture used to centralize page interactions
playwright-web-api-automation-ts/src/pages/notes-app.page.ts: page-object style selectors using data-testid and stable locator patterns
Test Coverage & Strategy
7/10
What and how to test
Clear test-coverage and risk strategy: unit+ui project separation, explicit coverage thresholds, tagged Playwright suites and numerous negative- and edge-case tests demonstrate a risk-aware strategy.
Evidence
wordwright/vitest.config.ts: two test projects (unit and ui) and strict coverage thresholds including 100% for src/engine/**
wordwright/src/engine/__tests__/reducer.test.ts: thorough unit tests covering edge cases and purity/idempotency
playwright-web-api-automation-ts/tests/api/notes-advanced.spec.ts: tests include negative paths, boundary checks and assertions annotated by priority tags
API & Integration Testing
7/10
Testing how parts work together
Solid API and integration testing: API client abstractions, schema checks, auth boundary tests and cross-layer UI/API validations show thoughtful contract and integration coverage.
Evidence
playwright-web-api-automation-ts/src/api/notes-api.client.ts: NotesApiClient abstraction for API interactions
playwright-web-api-automation-ts/tests/api/notes-crud.spec.ts: comprehensive CRUD integration tests using APIRequestContext
playwright-web-api-automation-ts/tests/api/notes-advanced.spec.ts: schema validation test (P7 — Response contract) validating response shape and types
Performance & Load Testing
5/10
Testing speed under load
Some performance awareness at the test level (response-time baselines and P95 checks) but no full load/soak harness in the analyzed human-authored files.
Evidence
playwright-web-api-automation-ts/tests/api/notes-advanced.spec.ts: P8 — Response time baseline with P95 calculation and thresholds
Bug Reporting & Analysis
3/10
Finding and describing bugs
Limited explicit bug-reporting artifacts; there is, however, evidence of investigation notes and local diagnostic comments in tests and sensible cleanup/teardown patterns but no linked OSS bug reports or root-cause writeups in the analyzed files.
Evidence
playwright-web-api-automation-ts/tests/notes-app/notes-app-ui.spec.ts: inline diagnostic comments describing ad-overlay flake diagnosis and mitigations
playwright-web-api-automation-ts/tests/advanced-ui/file-upload.spec.ts: test setup and cleanup with try/catch blocks showing pragmatic local debugging and fixtures
CI Test Integration
6/10
Running tests automatically
CI/test integration is present in test config and scripts (Allure outputs, reporters, coverage thresholds, retry/workers) but direct workflow orchestration files were in the non-scoring context rather than the human-authored set used for scoring.
Evidence
playwright-web-api-automation-ts/playwright.config.ts: reporter includes allure-playwright with outputFolder 'allure-results' and CI-friendly settings
playwright-web-api-automation-ts/package.json: scripts for test:regression, report:allure and setup automation
wordwright/vitest.config.ts: coverage thresholds, multiple test projects and setupFiles for jsdom UI tests
Expertise
SDET & Test Engineering• Senior
Unit & Component Testing• Middle
Technologies
QA
CI/CD
JavaScript
TypeScript• Senior
Rest API
Selenium
Playwright
Postman• 4y+
Allure Reporting
Pytest• 4y+
Pact
Vitest
Robot Framework• 4y+
Recommendations
- Lead development of cross-platform E2E and API automation suites, including fixture libraries and page-object modules (Playwright + APIRequest fixtures).
- Own migration or consolidation efforts from legacy Robot Framework suites into Playwright/TS automated pipelines while preserving negative-path and flake mitigations.
- Implement contract testing and schema-first validators (pact/schemathesis) to harden API contracts and add automated contract verification into CI.
- Design SLO-based test gates and integrate lightweight performance baselining (P95/P99 thresholds) into nightly runs with clear alerting/threshold logic
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
Middle Backend Developer
Confidence: Medium API Engineer
API-focused test automation and integration engineer (mid-level) with a strong emphasis on pragmatic API boundary testing and robust client-side validation. The strongest proven skill is end-to-end and contract-level API testing, demonstrated by comprehensive Playwright API suites and API client code (playwright-web-api-automation-ts/tests/api/notes-advanced.spec.ts and src/api/notes-api.client.ts). Public code does not show production backend service implementations, tuned database schema work, or large-scale distributed system design decisions.
API Design
4/10
How well APIs are designed
Consistent, pragmatic API boundary testing and contract checks are present, but there is little explicit API-versioning, idempotency key handling, or formal error contract design in service-code.
Evidence
playwright-web-api-automation-ts/tests/api/notes-advanced.spec.ts: response contract and auth-boundary tests
playwright-web-api-automation-ts/tests/api/notes-crud.spec.ts: core CRUD contract tests
playwright-web-api-automation-ts/src/api/notes-api.client.ts: API client interface and types
Data Layer & Database
3/10
Working with databases
Data-layer evidence is limited to client-side storage and runtime validation; there is careful input validation for persisted state but no server-side schema evolution or tuned SQL/DB work.
Evidence
wordwright/src/storage/schemas.ts: runtime validators for persisted storage and schema defaults
wordwright/src/storage/migrations.ts: migration hook present (schema evolution entrypoint)
wordwright/src/storage/sessionRepository.ts: session persistence and sanity checks
Scalability & Performance
2/10
Handling load and speed
Some performance measurement and baseline tests exist, but there is little evidence of caching strategy, queue-based decoupling, connection pooling, or measured scaling optimizations.
Evidence
playwright-web-api-automation-ts/tests/api/notes-advanced.spec.ts: response time baseline and P95 measurement
performance-focused pull request notes in NON-SCORING CONTEXT (not used for scoring) indicate attention to throughput in test suites
System Architecture
4/10
Overall system structure
Reasonable modular structure and separation of concerns for test automation and frontend storage, but no evidence of multi-service decomposition, inter-service contracts or large-scale orchestration decisions.
Evidence
playwright-web-api-automation-ts/src/pages/, src/services/, src/helpers/: clear module boundaries in the test framework
wordwright/docs/ARCHITECTURE.md and wordwright/src/storage/*: architecture and storage boundary design
Security & Auth
5/10
Protecting data and access
Clear security awareness at integration points (HMAC signing, secret loading, auth-boundary tests), though some operational secret handling and ACL choices could be improved.
Evidence
botium-chatbot-test-framework/connectors/whatsapp-webhook-connector.js: HMAC SHA-256 signature generation for webhook authentication
botium-chatbot-test-framework/auth.js: dotenv-based secrets loading with a production SSM scaffold
playwright-web-api-automation-ts/tests/api/notes-advanced.spec.ts: tests for invalid/expired tokens returning 401
Reliability & Observability
4/10
Stability and monitoring
Practical reliability patterns exist in tests and scripts (try/catch, cleanup, test teardown, defensive storage validation), but systematic retries/backoff, circuit breakers, and structured observability are limited.
Evidence
botium-chatbot-test-framework/python_send.py: try/except around execution, S3 upload and notification flow with error paths
playwright-web-api-automation-ts/tests/*: test cleanup via afterAll and cross-layer annotations
wordwright/src/storage/schemas.ts: defensive validation to avoid crashes from corrupt persisted data
Expertise
Messaging & Real-time• Middle
Microservices & API Architecture• Middle
Node.js• Middle
Technologies
Node JS• Middle
AWS
Dotenv
Recommendations
- Develop robust backend services and service-side API contracts, including versioning and idempotency handling for endpoints and long-running operations.
- Design and implement observable production patterns - structured logging with correlation ids, metrics (Prometheus) and alerting hooks, and retry/backoff strategies with reasoned timeouts.
- Lead integration work on webhook/message pipelines and real-time connectors, improving secrets management (SSM/External Secrets) and rotating credentials.
- Build small backend components (Node.js or Python) to solidify server-side data layer experience such as migrations, indexed queries, and transactional boundaries.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories: