Shawn Yap
Zheng Yi
Solution Architect, AI Systems
BeyondEdge · Singapore
Hello. I'm an AI engineer and data scientist. I ship production systems, from agent platforms and grounded retrieval to speech pipelines and classifiers, and measure the results. Email me at shawnyzy1@gmail.com. Recruiting? My resume is a one-page PDF.
Experience
AI engineering
- +13%reply fidelityProperty AI AssistantConsumer assistant across 9 intents, with tiered retrieval over 4 Qdrant collections and a LoRA fine-tuned model.
- In productionEnterprise agent platformMulti-tenant LLM agents on GKE, with model routing, per-team token budgets, and policy-gated MCP tools.
- Unsafe SQL blockedNL-to-SQL agentsSchema-aware agents with constrained decoding, query validation, and retries.
- 40% → 5%word error rateVoice pipelineASR–LLM–TTS in production, with tool routing, schema validation, and confidence gating. Whisper fine-tuned with LoRA.
please move the weekly review to nine on friday and send the notes to the hole team before noon today
Word error rate · Base Whisper 40%
Drag to compare. Illustrative sentence, real result from FCS Solutions.
Data science
- 5k+forecasts scoredSurvivorGenetic search over trading strategies, Brier-calibrated, with look-ahead bias ruled out across 676 backtest sessions.
- 30%EER improvedAudio deepfake detectionSpeech embeddings and prosodic features, on seed-fixed benchmarks with EER, ROC-AUC, and failure-mode analysis.
- 0.847accuracy, 0.781 F1POFMA actor classificationMulti-label BERT over 512-token chunks, with scraping, PDF pipelines, and contextual feature engineering.
- 200+classesIntent classificationFine-tuned classifier with calibrated confidence thresholds for production routing.
- Routedbook a room at nine
- Wrong routesend the notes to the hole team
- Fallbackcould you push friday back a bit
- Fallbackuh the thing from before
Coverage · Accuracy · Threshold
Drag to set the threshold. Illustrative requests, real technique from FCS Solutions.
Skills
- AI Engineering
- Python, Async Python, FastAPI, REST APIs, SQLAlchemy, PostgreSQL, Docker, Git, AWS, GCP, Azure, Pytest, Automated Testing, Model Deployment, Claude Code
- Data Science
- pandas, NumPy, SQL, Feature Engineering, Statistical Analysis, Experiment Design, Backtesting, Point-in-Time Evaluation
- Machine Learning
- PyTorch, Hugging Face Transformers, TensorFlow, LoRA/PEFT, Fine-tuning, BERT, Qwen, Whisper, NLP, ASR, Embeddings, Dataset Curation
- Agentic AI & LLM Systems
- Tool Calling, Tool Routing, MCP, RAG, Hybrid Retrieval, Qdrant, Structured Outputs, Constrained Decoding, NL-to-SQL, Prompt Engineering, LangChain, OpenClaw
- LLM Reliability & Evaluation
- Guardrails, Confidence Gating, Schema/Execution Validation, Retry Logic, Calibration, Threshold Tuning, WER, EER, F1, ROC-AUC, Brier Score, Error Analysis, Benchmarking
Education
- Nanyang Technological UniversityAug 2021 – May 2025Bachelor of Computing (Hons), Computer Science, Minor in Modern Languages
- University of StuttgartMar 2024 – Aug 2024Semester exchange and German immersion programme, Computer Science
- CertificateDeepLearning.AI TensorFlow Developer Specialization
- CertificateGoogle Advanced Data Analytics Certificate
- Singapore
- Latest push
- quantwachstum · Aug 2026
- Speaks
- English, Mandarin, Cantonese, German (B1)
- Learning
- German + Spanish