Software Engineer, Machine Learning Engineer (Agentic AI)

FuriosaAI · AI Transformation · Seoul, South Korea
공고 번호 4005371201

AI가 도구를 사용해 문제를 탐색하고, 실행 결과를 평가하며 개선하는 시스템 개발.

학사 명시 · 지원 완료

확인할 점: 접수 확인 메일 수신. RL·대규모 multi-agent 운영은 미확인.

영문 이력서 PDF ↗영문 이력서 Overleaf ZIP ↓

Cover Letter PDF ↗Cover Letter Overleaf ZIP ↓

이 공고의 지원서 열기 ↗

Career Summary

필수 Career Summary 칸에 붙여 넣기.

I am an ML engineer and AI Researcher at ISPARK R&D Center responsible for TieMing's core AI features, from model training to inference APIs and deployment. I bring a multimodal research background—five publications, four as first author—and turn experimental results into working products. I build AI workflows with validated tool execution, reusable context, and recoverable job state.

I built searchable knowledge and wiki workflows from selected coding sessions, with source links, incremental ingestion, and scoped access. I delivered installation tooling and developer-tool integration so others could adopt the workflow.

I developed a permission-aware CRM query workflow, improving schema retrieval and validating generated SQL before read-only execution. I designed and validated PostgreSQL Row Level Security (RLS) so data access is enforced independently of the model's output.

I bring direct experience building the parts an agent needs beyond generation: validated tool execution, reusable context, and recoverable jobs. At ISPARK, I develop TieMing's core AI features and take responsibility for their implementation through deployment. At FuriosaAI, I can contribute to agent execution and memory through permission-aware tools, traceable context, and explicit job state. I would use real workflows to validate how these components behave together, then turn recurring failure cases into focused improvements.

Cover Letter

PDF 첨부 또는 Enter manually에 붙여 넣기.

Dear FuriosaAI Hiring Team,

I am applying for the Software Engineer, Machine Learning Engineer (Agentic AI) position (Job ID 4005371201). I bring direct experience building the parts an agent needs beyond generation: validated tool execution, reusable context, and recoverable jobs. At ISPARK, I develop TieMing's core AI features and take responsibility for their implementation through deployment. Before ISPARK, I served one year as an information-systems NCO in the Republic of Korea Air Force (Dec 2021–Dec 2022), maintaining Linux servers through shell scripting and log analysis.

I built a system that turns selected coding-session records into searchable knowledge and reusable wikis. Engineering decisions were scattered across sessions, so I preserved links to original records and added incremental ingestion instead of repeatedly collecting the same history. Tenant-, department-, and user-level access controls determine which records are visible. I also delivered an installer and developer-tool integration, taking the work beyond a local prototype into a workflow another user could adopt.

I built a CRM query workflow covering schema retrieval, SQL validation, and controlled execution. I addressed incorrect table selection by enriching database context and combining semantic and keyword retrieval. I enforced read-only transactions and query limits, then designed and validated PostgreSQL Row Level Security (RLS) to restrict each user's row access. I also checked that document permission changes altered the evidence available to the answer. Execution checks and data access controls are part of the system, independently of the model's output.

At FuriosaAI, I can contribute to agent execution and memory through permission-aware tools, traceable context, and explicit job state. I would use real workflows to validate how these components behave together, then turn recurring failure cases into focused improvements.

Sincerely,
Jeonghun Kim