Bingxi Jiang
Software Developer
JavaScript
C++
SQL
Java
C
Node JS
TypeScript
Python
Data Layer & Database: 5/10
System Architecture: 5/10
API Design: 4/10
Active 1 day ago
Invite to interview
Message
Download CVCV
Overview
Technical skills
Timeline
Roles
Overview
Full-stack web developer (senior) specializing in real-time transcription UIs and LLM-powered note workflows. The strongest proven skill is integrating LLM-driven features from end to end as shown by the backend summarizer and Gemini integration in summarizer.js which powers concept detection, Q&A, and note generation. Public code does not show automated test coverage, CI pipelines, or formal load testing and benchmarking artifacts.
Technical skills
JavaScript• Senior
C++
SQL
Java
C
Node JS• Senior
TypeScript• Senior
Python• Senior
C++
PyTorch C++
Java
Spring Boot
Node JS
Express
Dotenv
BullMQ
Python
Pydantic
Uvicorn
FastAPI
Typer
Databases
PostgreSQL
Snowflake
Apache Kafka
MySQL
Redis
ElasticSearch
Frontend
Next.js
React.js
Vite
Vue.js
Tailwind CSS
Zod
AG Grid
DevOps
Amazon EC2
AWS
CI/CD
GCP
GitHub Actions
Nginx
Rest API
Yandex Cloud
Docker
Kubernetes
WebSockets
QA
Sentry
Playwright
AI/ML
AI Agents
Claude
Gemini
LangChain
LoRA
Multimodal AI
NumPy
PyTorch
RAG
Scikit-learn
Streamlit
PEFT
LLM
Pandas
Mobile
Twilio
Timeline
Software Engineer Intern
•
Junior
Dassault Systèmes
•
Internship
Engineered a closed-loop LLM pipeline using Playwright MCP to convert legacy Geb workflows into maintainable TypeScript test suites. Designed a modular TypeScript framework with reusable components including API handling, nested frames, and AG Grid for faster validation. Improved workflow runtime and increased the number of validation checkpoints while enabling idempotent reruns through isolated browser contexts and persisted workflow identifiers.
Playwright
TypeScript
AG Grid
LLM
Software Engineer Intern
•
Junior
GTechFin Inc
•
Internship
Built a Python/Typer workflow engine and an MCP service supporting rerunnable DAG execution for higher reliability. Modeled pipeline state in MySQL using transactions, indexes, and idempotent retries to improve consistency and reduce query latency. Benchmarked MongoDB and Elasticsearch, then implemented a self-hosted cosine index to accelerate video search and retrieval, improving overall video production turnaround time.
Typer
Python
MySQL
ElasticSearch
Software Engineer Intern
•
Junior
Fresh Road Inc
•
Internship
Developed a real-time Node.js/Express voice backend that streamed Twilio audio to Deepgram, LLM services, and CRM systems for concurrent calls. Implemented orchestration across multiple BullMQ/Redis queues with retries, backoff, deduplication, and recovery to improve reliability and queue latency. Reduced LLM cost and p95 latency via caching and context compression, and used Sentry with correlation IDs to trace HTTP and WebSocket flows for faster incident resolution.
Node JS
Twilio
BullMQ
Redis
WebSockets
LLMsince 2025
Sentry
Algorithm Engineer Intern
•
Junior
Ping An Technology
•
Internship
Created a six-stage Pandas and MySQL data pipeline for ingestion, quality control, classification, scoring, and reporting, cutting preprocessing time. Built an asynchronous FastAPI/WebSocket inference service with job tracking and structured outputs to return results within seconds. Optimized INT8 serving using dynamic batching to reduce p95 latency and increase throughput, and packaged the pipeline in Docker and Kubernetes for production-scale usage.
Pandas
MySQLsince 2025
FastAPI
WebSocketssince 2025
Docker
Kubernetes
Senior Backend Developer
Confidence: Medium API Engineer
A senior-level API-focused backend engineer specializing in real-time transcription and AI-driven note pipelines with a strength in integrating streaming audio, LLM backends, and a modular server/frontend stack. The strongest proven skill is building live transcription and summarization pipelines backed by LLM calls, shown concretely in summarizer.js (callGemini/generateNotes/extractActionItems) and server.js (WebSocket transcriber integration, maybeAutoSummarize). The public code does not show formal schema migration histories, production-grade observability (traces/metrics/alerting), or hardened operational policies like retry/circuit-breaker configurations for external LLM/ASR calls.
API Design
4/10
How well APIs are designed
Pragmatic REST + WebSocket API design with consistent error handling and timeouts at the client, but no formal versioning, idempotency keys, or pagination patterns and limited API-era contract/versioning discipline.
Evidence
Bingxi-Jiang/Noted/server.js: multiple REST endpoints (/api/sessions, /api/folders, /api/sessions/:id/notes/generate) and WebSocket /ws handling
Bingxi-Jiang/Noted/public/app.js: api(url, options) wrapper and withTimeout usage for client-side API timeouts
Bingxi-Jiang/Noted/test_functions/stage_dev/stage1/stage1_voice_rag_app.py: FastAPI endpoints (/api/transcribe, /api/ask) with clear request/response shapes
Data Layer & Database
5/10
Working with databases
Conventional relational storage with conscious design choices (indexes, WAL, chunked transcripts) and transactional patterns, but no migration chain or advanced schema-evolution tooling exposed.
Evidence
Bingxi-Jiang/Noted/test_functions/stage_dev/stage1/stage1_voice_rag_app.py: init_db() creates transcripts table and index idx_transcripts_session_chunk
Bingxi-Jiang/Noted/test_functions/stage_dev/stage1/stage1_voice_rag_app.py: get_conn() sets PRAGMA journal_mode=WAL and uses connection context managers for commits
Bingxi-Jiang/Noted/server.js: uses db.js functions (insertChunk, getAllChunks, getRecentChunks) to manage transcript/summaries
Scalability & Performance
4/10
Handling load and speed
Design shows attention to streaming performance and UI responsiveness (audio chunking, downsampling, transcript preview limits) and background work, but lacks demonstrated system-level caching, deliberate load-testing, or queue-based decoupling configs.
Evidence
Bingxi-Jiang/Noted/public/app.js: downsampleFloatToInt16 and chunked audio streaming to WebSocket; transcript preview truncation to keep UI responsive
Bingxi-Jiang/Noted/server.js: WebSocket-based transcriber integration and live broadcastToSession for streaming transcript updates
Bingxi-Jiang/Noted/artified_backend/serve.py: CaptureManager with background thread and configurable capture interval
System Architecture
5/10
Overall system structure
Clear module boundaries and a purposeful decomposition (db, summarizer, exporter, transcriber, server, front-end) with background workers and scheduler, demonstrating deliberate architecture for an integrated real-time+batch product.
Evidence
Bingxi-Jiang/Noted/server.js: imports and composes db.js, summarizer.js, transcriber.js and exporter.js to shape distinct responsibilities
Bingxi-Jiang/Noted/exporter.js: separate DOCX/PDF exporter module encapsulating export logic
Bingxi-Jiang/Artified-Self/artified_backend/serve.py: separation of capture, build pipelines, Google OAuth, and web API endpoints
Security & Auth
4/10
Protecting data and access
Reasonable security hygiene in places (API-key gating, OAuth state check, input sanitization on UI), but CORS wide-open in dev code and limited defensive measures against injections or strong threat modeling across interfaces.
Evidence
Bingxi-Jiang/Noted/server.js: early checks for GEMINI_API_KEY and DEEPGRAM_API_KEY and guarded API error returns
Bingxi-Jiang/Noted/public/app.js: escapeHtml, escapeAttr, sanitizeFilename utilities used before rendering or file/export operations
Bingxi-Jiang/Artified-Self/artified_backend/serve.py: google_auth_callback validates OAuth state
Reliability & Observability
4/10
Stability and monitoring
Practical reliability features (WS error handlers, background scheduler, graceful capture pause/resume) and client timeouts, but limited evidence of advanced resilience patterns like retries with backoff, circuit breakers, or structured distributed tracing/metrics.
Evidence
Bingxi-Jiang/Noted/public/app.js: withTimeout wrapper for client requests and WS.onerror / onclose handling in openSessionSocket/handleSocketMessage
Bingxi-Jiang/Noted/server.js: transcriber.on('error') handling and maybeAutoSummarize guard logic (debounce/checks) to avoid repeated calls
Bingxi-Jiang/Artified-Self/artified_backend/serve.py: scheduler startup/shutdown hooks and CaptureManager pause/resume semantics
Expertise
Backend AI & LLM• Senior
Databases & Vector Storage• Middle
Messaging & Real-time• Senior
Microservices & API Architecture• Senior
Node.js• Senior
Python• Senior
System Architecture• Senior
Industries
Education• Senior
Lifestyle• Senior
Professional Services• Senior
Technologies
Java
Node JS• Senior
Rest API
Express
Spring Boot
FastAPI
Apache Kafka
Pydantic
Uvicorn
Dotenv
Stack• mentioned only
Recommendations
- Extend and harden operational tooling: add structured logging, request tracing, metrics and well-documented retry/backoff policies around external AI/ASR calls.
- Drive schema evolution and migrations: add a migration history/tooling for the DB (not just init scripts) and document backward-compatible migration strategy.
- Build load and resilience test artifacts: simulate heavy concurrent WS/audio ingest and LLM failures to validate autoscaling and backpressure strategies.
- Package LLM/AI call wrappers with explicit timeouts, retry with jitter, and graceful degradation fallbacks to reduce token spend and avoid blocking user flows.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
Senior Frontend Developer
Confidence: Medium Fullstack
Full-stack web developer (senior) specializing in real-time transcription UIs and LLM-powered note workflows. The strongest proven skill is integrating LLM-driven features from end to end as shown by the backend summarizer and Gemini integration in summarizer.js which powers concept detection, Q&A, and note generation. Public code does not show automated test coverage, CI pipelines, or formal load testing and benchmarking artifacts.
UI Component Architecture
5/10
How interface parts are built
Modular, function-based UI with deliberate render boundaries and tokenized CSS, but not a component-library-based architecture; clear separation of render helpers, but no custom design-system package or React component model.
Responsive & Cross-browser
5/10
Works on all screens and browsers
Responsive layout with breakpoints, dark theme, and progressive collapse rules. Uses feature-detection around media devices and provides graceful fallback messages, but no explicit RTL or full i18n scaffolding.
Performance Optimization
6/10
Speed of the interface
Measured performance-minded patterns: rAF batching, transcript preview truncation to avoid rendering huge lists, audio downsampling in client, and binary WebSocket audio streaming. No published before/after metrics or CI bundle analysis artifacts were found.
Accessibility & Semantics
6/10
Usable for everyone
Accessibility and keyboard/focus discipline are intentionally handled: focus capture/restore, keyboard escape handling, and ARIA roles on custom widgets. Evidence is implementation-level rather than relying on an external UI-kit.
State Management & Data Flow
6/10
Managing data in the app
Robust client-server state and async handling: single global state object for UI, explicit recording cache sync, concurrency guard tokens for session loading, timeouts for API calls, and deliberate polling cancellation. No evidence of a full formal state machine library, but careful ad-hoc async discipline is present.
UX & Visual Polish
6/10
Look and feel quality
Good UX polish across many edge states: empty states, skeleton partial transcript, progress panels, toasts, and export modals. The app balances perceived performance and usability, but lacks explicit undo patterns or measured RUM metrics.
Expertise
Frontend AI Integration• Middle
React• Junior
Industries
Education• Middle
Technologies
Vue.js
Tailwind CSS
Next.js
Yandex Cloud
React.js
Vite
BullMQ
Zod
AG Grid
Recommendations
- Build real-time meeting capture and collaboration features that rely on audio/screen capture, WebSocket streaming, and LLM summarization.
- Develop LLM-powered knowledge workflows such as RAG Q&A, cross-session search, and export pipelines (PDF/DOCX) where you can leverage the existing summarizer and exporter modules.
- Extend the client to a component-based framework or design system (React/Next) and add automated tests and performance benchmarks to harden the product for production.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
Senior QA Engineer
Confidence: Medium Generalist
Senior full-stack engineer specializing in real-time audio transcription and AI-powered note generation. The strongest proven skill is building an end-to-end realtime transcription pipeline and interactive UI backed by a WebSocket server and AI summarization logic, evidenced by public/app.js and server.js plus the summarizer.js modules. There is limited evidence of formal automated test frameworks, CI integration, or performance/load testing in public code.
Test Automation Frameworks
2/10
Building automated tests
Small collection of bespoke test scripts and QA helpers for audio/transcript flows, but no structured test framework, fixtures, or parallelization.
Test Coverage & Strategy
2/10
What and how to test
Some focused analysis and provider comparison scripts exist for negative-path and multi-provider probing, but no property-based tests, mutation testing, or formal risk-based tagging structure.
API & Integration Testing
3/10
Testing how parts work together
API endpoints are exercised by lightweight integration scripts and a FastAPI stage app; there is deliberate error handling and context-aware retrieval logic but no contract testing tools or schema-driven tests.
Performance & Load Testing
Testing speed under load
Not evidenced in public code
Bug Reporting & Analysis
1/10
Finding and describing bugs
Basic runtime logging and try/catch error reporting are present, but no formal bug reports, issue triage documents, or postmortem artifacts.
CI Test Integration
Running tests automatically
Not evidenced in public code
Expertise
API Testing & Contract Validation• Middle
SDET & Test Engineering• Junior
Industries
Blockchain & Crypto• Middle
Technologies
Python• Senior
JavaScript• Senior
TypeScript• Senior
Playwright
WebSockets
Sentry
Recommendations
- Develop realtime transcription and note-taking features that integrate WebSocket audio streaming, server-side diarization, and LLM-based summarization.
- Build API-first integrations and contract tests around the QA endpoints, including schema validators and schemathesis or contract tests to harden RAG and Q&A endpoints.
- Implement a minimal CI pipeline with selective test runs and failure artifacts for browser and API tests to accelerate delivery and reduce regression risk.
- Prototype a scalable deployment for the transcriber components including containerization and basic autoscaling to validate production behavior under load.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories: