Eric H.

Key Skills

Software

Don't disclose

Top Subject Matter

LLM distillation

Embeddings Domain Expertise

agentic retrieval for customer fit

Top Data Types

Text

Image

Top Task Types

Fine-tuning

Diagnosis

Data Collection

Freelancer Overview

Machine Learning Research Intern at Apta (LLM distillation for product/customer fit embeddings). Brings 4+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Don't disclose and PyTorch. Education includes Master of Engineering, University of Cambridge (2026) and High School Diploma, Caulfield Grammar School (2021). AI-training focus includes data types such as Text, Medical, and DICOM and labeling workflows including Fine-tuning, Diagnosis, and Data Collection.

Labeling Experience

GPU Hardware Intern at Imagination Technologies

Don't discloseData Collection

Worked on GPU hardware as an intern with relevance to machine learning training workloads. Although labeling is not explicitly stated, the role supported AI training infrastructure and performance needs for ML systems. The position likely contributed to optimizing compute for downstream model training and experimentation. • GPU hardware integration/optimization for ML training workflows • Performance considerations for training workloads • Engineering support for AI experimentation hardware • Collaboration within a hardware-focused ML ecosystem

2025 - 2025

Machine Learning Research Intern at Apta (LLM distillation for product/customer fit embeddings)

Don't discloseTextFine-tuning

Developed a new model architecture and training technique to distill LLM understanding into an embedding model by self-selecting training data autoregressively. Built supporting agentic systems to perform intelligent database querying and custom tool calling to support a luxury shopping chat bot. The work involved preparing and curating training data and optimizing training flows for downstream embedding use. • Training-data selection and autoregressive self-selection for distillation • Agentic tool-calling and database querying to maintain customer personality context • Implemented/iterated model training methodology for efficient LLM distillation • Prepared research outputs (e.g., preprint) describing the training approach

2025 - 2025

Machine Learning Research Intern at University of Cambridge (seizure detection CV models)

Diagnosis

Developed novel computer vision model architectures in PyTorch for detecting seizures in infant ICU settings. The project required preparing labeled visual data and training/evaluating detection models for medical diagnosis use cases. Coordination with a small team included strategy to test ideas and iterate on model performance. • Seizure detection model development using PyTorch • Training/evaluation on medical image data for diagnostic detection • Team-based iteration of model architectures and activations • Version control and collaborative experimentation using Git

2024 - 2024

Education

Work History