ILHAM HAKIM
I’m Ilham Hakim, an AI engineer with 4+ years of hands-on experience across machine learning, LLM post-training, and self-hosted agentic infrastructure.
About
I’m Ilham Hakim, an AI engineer with 4+ years of hands-on experience across machine learning, LLM post-training, and self-hosted agentic infrastructure. I build production-ready, local-first AI systems—from sovereign training and reproducible course infrastructure to scalable Kubernetes deployments that serve models securely.
I also enjoy bridging research-grade intelligence into practical tools: I’ve architected RAG “second brain” systems with semantic recall, and I’ve led LLM post-training and quality calibration using SFT and RLHF workflows. I thrive across the full stack of AI delivery—data pipelines, orchestration, access control, evaluation, and deployment—so teams can move from prototypes to reliable real-world outcomes.
- 3 years' experience
- Joined September 2026
Languages
- IndonesianFluent
- EnglishAdvanced
- Spanish; CastilianBeginner
Industry experience
- Computers & Electronics
- Education
- Financial Services
- Software & Internet
Work experience· 4 roles
Led curriculum design and instruction for Indonesia's sovereign AI training program; built reproducible course infrastructure on a local-first stack (GitLab, Kubernetes, skill-based onboarding). Deployed sovereign AI infrastructure on production Kubernetes, serving GLM 5.2 via LiteLLM gateway with GitLab and Zitadel OIDC for access control. Architected a self-hosted RAG second-brain system using Docker and Postgres+pgvector, with a custom Python MCP server bridging mem0 to opencode for cross-session semantic recall.
Researched equities, commodities, and FX opportunities across the family office portfolio, blending technical, fundamental, and quantitative analysis to support investment decisions. Backtested trading strategies in Python (pandas, Google Colab) to validate hypotheses and surface actionable portfolio signals. Built Medallion, a multi-platform market intelligence pipeline using faster-whisper transcription and LLM-synthesized digests to track Indonesian ticker sentiment.
Conducted post-training of LLMs across 12 campaigns via SFT and RLHF workflows, developing chain-of-thought reasoning, multimodal understanding and generation, tool use and browsing, memory/context handling, code execution, and multi-step planning capabilities. Increased calibration alignment from 70% to 90% across 26 Quality Analysts within one month by building an internal knowledge hub consolidating guidelines, examples, and updates into a single source of truth.
Education
Certifications and awards
Hire ILHAM HAKIM today
Post a job and invite ILHAM, or contact them directly. No hidden commission.