Taehyun Lee
Taehyun Lee

Taehyun Lee

Backend • Middle • 29 years • Seattle, USA • UTC-08:00 • Relocation • Work permits: United States
Software Developer
4+ years exp
Java
Kotlin
TypeScript
JavaScript
Python
Go
API Design: 4/10
Active 2 days ago
Invite to interview
Message
Download CVCV
Overview
Technical skills
Timeline
Roles

Overview

A backend-focused API engineer (Middle) who builds Python FastAPI services with relational schemas and migration workflows; their strongest area is building REST endpoints and CRUD-backed database logic. The most proven skill is relational application development using SQLAlchemy/Alembic and hand-crafted SQL, demonstrated in migrations (migrations/versions/5e53e203bfc3_create_tables.py) and extensive CRUD logic (src/crud/user.py, src/crud/activity.py). There is limited evidence of production-grade resilience, secure query parameterization (many f-string SQL queries), observability, or advanced distributed-system design in public code.

Technical skills

Java
Kotlin
TypeScript
JavaScript
Python• Middle • 4y+
Go• Middle • 3y+
Java
Spring Boot
Python
FastAPI
SQLAlchemy
Alembic
Boto3
Pydantic
Uvicorn
Databases
FAISS
MySQL
TimescaleDB
Apache Kafka• 3y+
DynamoDB• 3y+
PostgreSQL• 3y+
DevOps
Datadog
Docker
Git
Grafana
Kubernetes
Loki
Prometheus
AWS• 3y+
GCP• 3y+
GitHub Actions• 3y+
gRPC• 3y+
Rest API• 3y+
QA
Sentry
AI/ML
LLM• 3y+
RAG• 3y+

Timeline

University of Washington
Master's Degree • Electrical & Computer Engineering
2026–2028 Seattle, Washington
Software Engineer II • Middle
KARROT • Full-Time
Jan 2026 to Jun 2026 5 Months Seoul In office
  • Built a Go-based internal observability platform using a microservice dependency graph to identify affected services during production incidents in under 2 minutes, reducing manual incident triage across 10+ engineering teams
  • Migrated time-series data from DynamoDB to PostgreSQL (TimescaleDB), using compression to reduce storage usage, cutting p50 aggregation API latency from ~1 minute to sub-second, validating scalability and performance on 100M+ records
  • Designed and built a Kafka-based event-driven multi-cloud video-processing pipeline across AWS and GCP to reduce single-provider dependency and improve resilience, processing 300,000+ videos per day
Go
DynamoDB
PostgreSQL
Apache Kafka
AWS
GCP
Hankuk University of Foreign Studies
Bachelor's Degree • Artificial Intelligence Convergence (Software & AI)
2016–2025 Seoul, South Korea
Software Engineer I • Middle
KARROT • Full-Time
Mar 2023 to Dec 2025 2 Years 9 Months Seoul In office
  • Built a shared backend video-processing and serving platform with standardized HTTP and gRPC APIs, consolidating fragmented pipelines across 3 engineering teams and expanding adoption to 8 teams
  • Contributed 4 upstream pull requests to an open-source feature-flag platform and deployed it to replace a legacy system that could only turn features on or off for all users at once, driving adoption across 6 engineering teams and enabling staged rollouts
  • Built a GitHub Actions deployment pipeline for a Go-based image-processing application across 8 regions, reducing deployment time by over 85%, from 20+ minutes to under 3 minutes
gRPC
Rest API
Gosince 2023
GitHub Actions
Software Engineer Intern • Junior
KARROT • Internship
Jan 2023 to Jan 2023 0 Months Seoul In office
  • Replaced a third-party meeting playback SaaS limited by poor Korean transcription with a Python-based internal service providing Korean transcription and multilingual captions, reducing vendor costs by 90%+
  • Prototyped a Python-based LLM-powered Q&A feature for past meetings using RAG, enabling employees to find information from past meetings without manually reviewing full recordings
Python
LLM
RAG
Middle Backend Developer Confidence: High API Engineer
A backend-focused API engineer (Middle) who builds Python FastAPI services with relational schemas and migration workflows; their strongest area is building REST endpoints and CRUD-backed database logic. The most proven skill is relational application development using SQLAlchemy/Alembic and hand-crafted SQL, demonstrated in migrations (migrations/versions/5e53e203bfc3_create_tables.py) and extensive CRUD logic (src/crud/user.py, src/crud/activity.py). There is limited evidence of production-grade resilience, secure query parameterization (many f-string SQL queries), observability, or advanced distributed-system design in public code.
API Design
4/10
How well APIs are designed
Reasonable REST API structure with FastAPI routers, pydantic schemas and response examples but lacking consistent error contracts, versioning beyond v1 and advanced API resilience/idempotency patterns.
Evidence
kids-diary-kida/kida-backend/src/api/v1/user.py - FastAPI routes, dependency-based auth and response handling kids-diary-kida/kida-backend/src/schema/user/response_example/likes.py - explicit response examples for endpoints
Data Layer & Database
3/10
Working with databases
Clear relational schema and Alembic migrations with SQLAlchemy models, but heavy use of hand-crafted raw SQL via f-strings without parameterization and limited transaction management reduces robustness.
Evidence
kids-diary-kida/kida-backend/migrations/versions/5e53e203bfc3_create_tables.py - Alembic migration creating schema kids-diary-kida/kida-backend/src/crud/user.py - extensive raw SQL queries built with f-strings and many db.commit() calls kids-diary-kida/kida-backend/src/database/session.py - engine and sessionmaker usage (DB layer wiring)
Scalability & Performance
2/10
Handling load and speed
Some attention to bulk ingestion and SQL-side grouping but missing explicit caching, queueing, connection pooling tuning or measured performance optimizations.
Evidence
kids-diary-kida/kida-backend/src/crud/activity.py - bulk_create implementation for ingesting activities kids-diary-kida/kida-backend/src/crud/user.py - complex SQL (GROUP_CONCAT, CTEs) used for query-side aggregation
System Architecture
3/10
Overall system structure
Reasonable project modularization (api, crud, model, schema, util, core) and pydantic settings, but single-service monolith style and no advanced service decomposition, secrets/config delivery or graceful degradation patterns shown.
Evidence
kids-diary-kida/kida-backend/src/core/app_config.py - pydantic Settings and get_settings pattern kids-diary-kida/kida-backend/src/api/v1/user.py and src/crud/* - clear folder separation for API, CRUD and models
Security & Auth
3/10
Protecting data and access
Auth and crypto basics are implemented (bcrypt via passlib, JWT tokens), but input boundary protections are incomplete due to raw SQL construction and there is no evidence of dependency vulnerability auditing or token lifecycle beyond issuing access tokens.
Evidence
kids-diary-kida/kida-backend/src/crud/user.py - password hashing with passlib and jwt token creation kids-diary-kida/kida-backend/src/util/auth.py - auth_user util (JWT validation) referenced by routes
Reliability & Observability
2/10
Stability and monitoring
Basic error handling in endpoints and a Gunicorn config exist but advanced reliability and observability patterns (timeouts, retries with backoff, structured logs with correlation ids, metrics/alerts) are not present.
Evidence
kids-diary-kida/kida-backend/src/api/v1/user.py - try/except blocks returning HTTP 500 with error text kids-diary-kida/kida-backend/src/core/gunicorn_config.py - basic worker configuration
Verified artifacts
Unleash/unleash• ★ 13.8K pydantic/FastUI• ★ 9K leanprover-community/lean4game• ★ 544 AUSG/anna-v2• ★ 6
Expertise
Python• Middle
Databases & Vector Storage• Middle
Microservices & API Architecture• Middle
Technologies
Python• Middle • 4y+
Go• Middle • 3y+
Java
Kotlin
MySQL
PostgreSQL• 3y+
Rest API• 3y+
gRPC• 3y+
GCP• 3y+
DynamoDB• 3y+
FAISS
Spring Boot
TimescaleDB
SQLAlchemy
GitHub Actions• 3y+
Loki
Datadog
FastAPI
Prometheus
Git
AWS• 3y+
Docker
Kubernetes
Apache Kafka• 3y+
Grafana
Pydantic
Uvicorn
Boto3
Alembic
Recommendations
  • Use parameterized queries or SQLAlchemy Core/ORM methods instead of string interpolation to eliminate SQL injection risk; refactor queries in src/crud/* to use bound parameters.
  • Consolidate multi-statement operations into explicit transactions with rollback on failure to ensure consistency (replace repeated db.commit() calls in src/crud/* with transactional context managers).
  • Add structured logging, request correlation ids and basic metrics (Prometheus/OpenTelemetry) and implement timeouts/retries where external I/O is used (e.g., boto3, external requests in src/util/open_api.py).
  • Introduce input validation and stricter error contracts for APIs (consistent error response format) and add tests covering auth flows and invitation-code edge cases.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
0417taehyun kida-backend db-db-deep