Ethan Gopez
Software Engineering Intern
Management: Under a year
Python
C++
SQL
Java
TypeScript
JavaScript
C#
C
Model Architecture & Training: 5/10
Data Pipeline & Feature Engineering: 4/10
Active 29 days ago
Invite to interview
Download CVCV
Overview
Technical skills
Timeline
Roles
Overview
A machine-learning focused developer (middle level) producing self-contained model, dataset, and training code with practical devcontainer tooling; the single most important strength is shipping runnable end-to-end training code. The strongest proven skill is applied Python ML engineering as shown by the custom ResNet, Dataset, and training loop in model/model.py, model/load.py, and model/train.py. There is limited evidence of production-grade CI/CD, infrastructure-as-code, observability, or cloud cost and security hardening in public code.
Technical skills
Languages
8
Python
C++
SQL
Java
TypeScript
JavaScript
C#
C
AI/ML
10
OpenCV
PyTorch
MediaPipe
Machine Learning
NumPy
Pandas
Scikit-learn
LLM
Gemini
AWS Bedrock
Other
28
Cloudflare
AWS
Matplotlib
React.js
ASP.NET Core
ROS2
ROS
Unity
Oracle
Playwright
Qiskit
PyTorch C++
Svelte
GitHub
Git
Linux
Rest API
PyGame
Claude Code
Gazebo
AI Agents
AI/ML
Deep Learning
Containers
CI/CD
Computer Vision
Human-in-the-Loop
Reinforcement Learning
Timeline
Undergraduate Researcher
•
Middle
Autonomous Mobile Robotics Laboratory (UT Austin)
•
Full-Time
Developing an LLM-driven reinforcement learning framework focused on human-feedback meta-optimization. Building core Python infrastructure and baseline training pipelines to support research toward publication. Applying research-grade software practices to iterate on training and evaluation workflows.
LLM
Python
Reinforcement Learning
Isaac Lab
Tech Lead
•
Lead
ECLAIR (Engineering and Computational Learning of AI in Robotics)
•
Full-Time
Leads the design of a robotic arm interface that uses EMG-based motion control, 3D vision for position tracking, ROS2 Gazebo simulation, and haptic feedback. Writes C++ code and connects supporting hardware via Arduino circuitry. Integrates an open-source AWS DeepRacer-based reinforcement learning approach for autonomous driving. (student organization)
C++
AWS
Machine Learning
ROS2
Reinforcement Learning
Computer Vision
MediaPipe
University of Texas at Austin
Bachelor's Degree •
Computer Science (Minor in Robotics)
Software Engineering Intern
•
Junior
Cox Automotive Inc.
•
Internship
Developed a full-stack reporting feature using ASP.NET Core and a front-end templating approach to support large exports and bulk edits backed by Oracle. Owned an end-to-end Playwright automation framework for production authentication flows, including CI/CD and team notifications. Overhauled a .NET API regression suite, expanding coverage across many REST endpoints and helping surface production defects. Prototyped an AI agent using AWS Bedrock for human-in-the-loop recommendation workflows.
ASP.NET Core
Oracle
Playwright
CI/CD
Slack
AWS Bedrock
Rest API
Machine Learning Programmer
•
Middle
Longhorn Neurotech
•
Full-Time
Developed and optimized an EEG-processing machine learning model in PyTorch to classify six degrees of freedom in virtual space. Experimented with convolutional neural network and capsule network architectures using hyperparameter tuning. Improved prediction accuracy on a frequency dataset through loss-function experimentation.
Python
PyTorch
Machine Learning
Middle AI/ML Engineer
Confidence: Medium ML Engineer
Image-focused ML engineer (mid-level) who implements end-to-end model building and training for small-scale classification tasks. Strongest proven skill is building and training a custom PyTorch ResNet-style model and data pipeline as shown by model/model.py, model/load.py and model/train.py. Lacks production-grade MLOps, rigorous experiment tracking, large-scale performance engineering and thorough evaluation tooling in public code.
Model Architecture & Training
5/10
How well models are designed and trained
Custom model and training loop for a simple residual network with reasonable ML practices (device handling, optimizer choice, loss).
Evidence
Simple-ResNet-Classifier/model/model.py: ResNet and ResNetBlock class definitions implementing custom residual blocks
Simple-ResNet-Classifier/model/train.py: training loop using AdamW, CrossEntropyLoss, epoch/validation loop and model checkpointing
Data Pipeline & Feature Engineering
4/10
How data is prepared for models
A minimal but correct PyTorch Dataset/DataLoader implementation and basic feature-transformation utilities (PCA) for data prep.
Evidence
Simple-ResNet-Classifier/model/load.py: TumorImageDataset and load_data returning a DataLoader
QAID/data_trans.py: pca_data_transformation implementing PCA-based feature reduction
Experimentation & Evaluation
3/10
How results are measured and tested
Basic experimentation practices: per-epoch metric logging, best-model saving, and simple train/validation splitting; no experiment-tracking system or rigorous eval suite.
Evidence
Simple-ResNet-Classifier/model/train.py: epoch prints, epoch_train/val accuracy aggregation and torch.save of best model
QAID/svm.py: joblib.dump of trained SVC and printed scoring
MLOps & Deployment
2/10
How models are shipped to production
Minimal MLOps footprint limited to saving model artifacts; no serving, CI integration, versioning or monitoring present.
Evidence
Simple-ResNet-Classifier/model/train.py: torch.save(model.state_dict(), 'model/resnet.th')
QAID/svm.py: joblib.dump(clf, model_name)
Computational Efficiency
2/10
How efficiently computing resources are used
Some attention to runtime (CUDA detection, pin_memory, configurable num_workers) but no profiling, distributed training, or quantization/optimizations.
Evidence
Simple-ResNet-Classifier/model/train.py: torch.cuda.is_available() handling and pin_memory usage
Simple-ResNet-Classifier/model/load.py and load_data: exposes num_workers and pin_memory parameters
Research Depth & Innovation
2/10
Depth of research and new ideas
Surface-level research artifacts (quantum ML experiment snippets and SPSA usage) but primarily exploratory and not full paper reproductions or rigorous ablations.
Evidence
QAID/spsa.py: use of SPSA, pauli_two_design and lambeq CircuitParser with explanatory comments
QAID/qsvm.py and QAID/svm.py: quantum/classical SVM pipelines and PCA-based feature preparation
Expertise
Computer Vision & Image Analysis• Middle
Medical AI & Healthcare• Middle
Industries
Artificial Intelligence• Middle
Health Care• Middle
Technologies
AI/ML
SQL
C++• since 2024 • Junior
Oracle• since 2026
OpenCV• since 2024
Computer Vision• since 2026
AI Agents
AWS Bedrock• since 2026
CI/CD• since 2026
Git
PyTorch• since 2024
AWS• since 2026
Cloudflare• since 2025
Gemini
LLM• since 2026
MediaPipe• since 2024
PyTorch C++
GitHub
Human-in-the-Loop• since 2026
Linux
Recommendations
- Develop production-ready MLOps flows: add reproducible experiment tracking (W&B/MLflow), clear config management, and CI checks for training and model validation.
- Improve evaluation rigor: add held-out test harnesses, metric logging, and automated ablation/validation scripts for reproducibility.
- Work on computational efficiency and scalability: add profiling, mixed-precision/AMP training, and basic quantization or batching experiments.
- Expand model architectures toward conv-based ResNets for image inputs and include unit tests for Dataset/Preprocessing code to increase reliability.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
Middle Data Scientist
Confidence: Medium ML Practitioner
A middle-level ML practitioner who builds small-scale deep learning and classical ML prototypes with a clear strength in end-to-end model code and experiment scripts. The strongest proven skill is implementing and training prototype models, evidenced by the custom ResNet, Dataset/DataLoader and training loop with validation and checkpointing in model/model.py and model/train.py. There is limited evidence of production hardening, rigorous statistical evaluation, experiment reproducibility tooling or deployment considerations in public code.
Statistical Rigor
2/10
Correct use of statistics
Minimal statistical rigor: basic train/validation accuracy reporting and a single train/test split in another script, but no uncertainty quantification, significance testing, multiple-comparison control or honest negative-result discussion.
Data Wrangling & Cleaning
4/10
Preparing and cleaning data
Practical data-wrangling for ML pipelines is implemented: custom Dataset/DataLoader, basic transforms, CSV label handling and PCA-based transformation are present, but there is little evidence of robust missing-value handling, provenance tracking or leakage checks.
Exploratory Analysis & Visualization
2/10
Exploring and visualizing data
Exploratory analysis is minimal and largely ad hoc: a few printed PCA outputs and diagram drawing, but no structured EDA, annotated findings or visual storytelling tied to actionable hypotheses.
Predictive Modeling
4/10
Building models that predict
Predictive modeling competence is clear at a prototype level: a custom simple ResNet-like model and training loop with validation and checkpointing exist, plus classical SVM pipelines for experiments; however cross-validation, calibration, detailed error analysis and production concerns are not present.
Business Insight & Impact
1/10
Turning analysis into business value
Very little evidence of business framing or impact analysis; the README states a simulated tumor-classification goal but there is no cost-of-error analysis, stakeholder metrics or deployment/operational reasoning.
Reproducibility & Notebook Hygiene
3/10
Clean, repeatable analysis
Basic reproducibility practices exist such as a requirements.txt, argparse entrypoint and explicit random seeds, but there is no evidence of CI, data/versioning (DVC), pinned environments beyond requirements, or structured experiment tracking.
Expertise
Data Science• Middle
Industries
Health Care• Middle
Technologies
Deep Learning
Python• since 2024 • Middle
Scikit-learn
Matplotlib
Pandas
NumPy
Machine Learning• since 2025
Recommendations
- Develop small-to-medium ML prototypes and research proofs-of-concept where quick iteration and model experimentation are required.
- Extend current work into robust pipelines: add cross-validation, detailed error analysis, calibration checks and automated experiment tracking (W&B/MLflow) for reproducibility.
- Hardening and productionization tasks such as creating CI tests, data/versioning (DVC), containerized training/inference and explicit data validation should be assigned to ensure deployability.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
Middle DevOps Engineer
Confidence: Medium Generalist
A machine-learning focused developer (middle level) producing self-contained model, dataset, and training code with practical devcontainer tooling; the single most important strength is shipping runnable end-to-end training code. The strongest proven skill is applied Python ML engineering as shown by the custom ResNet, Dataset, and training loop in model/model.py, model/load.py, and model/train.py. There is limited evidence of production-grade CI/CD, infrastructure-as-code, observability, or cloud cost and security hardening in public code.
CI/CD Pipelines
Automated build and deploy
Not evidenced in public code
Infrastructure as Code
Managing servers with code
Not evidenced in public code
Containerization & Orchestration
2/10
Working with containers
Minimal containerization evidence focused on a development container startup script that shows careful local-only VNC setup and process checks but lacks build, multi-stage optimization, non-root/container-hardening, or orchestration manifests.
Observability & Monitoring
Watching system health
Not evidenced in public code
Reliability & Incident Response
1/10
Keeping systems up
Small-scale reliability practices are present in scripts and training code (stale X-server lock handling and simple checkpointing) but there are no deploy strategies, runbooks, incident postmortems, or DR rehearsal artifacts.
Cloud & Cost Optimization
Smart use of the cloud
Not evidenced in public code
Verified artifacts
Expertise
Platform Engineering & IDP• Junior
Industries
Artificial Intelligence• Middle
Health Care• Middle
Technologies
Containers
Recommendations
- Develop small ML prototypes and experiments, including model definition, dataset pipelines, and training loops for classification tasks.
- Build reproducible training infra by adding versioned checkpoints, artifact uploads, and a simple CI job to validate training runs.
- Harden development containers and deployment artifacts by adding Dockerfiles, multi-stage builds, non-root container user, and image scanning.
- Implement basic observability and reliability practices such as training-run logging, artifact immutability, and simple alerting or monitoring hooks.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories: