Overview
Aquila is Eagle Point AI's internal platform for producing and evaluating software-engineering training data. I work inside it daily, building coding tasks, reviewing other contributors' datasets, and rewriting prompts until every constraint is preserved.
Contributions
Coding dataset production
Author Python, JavaScript, SQL, and debugging tasks aligned with LLM fine-tuning specs. Structure prompts so models learn correct patterns, not shortcuts or ambiguous edge cases.
- Python
- JavaScript
- SQL
- LLM Fine-tuning
Quality review & consistency
Review datasets from other developers against Eagle Point standards. Catch logical errors, ambiguous specs, and constraint drift before data enters training pipelines.
- Quality Review
- Dataset Validation
- Peer Review
Prompt specification & rewrites
Rewrite prompts to match exact requirements without dropping a single constraint. That kind of detail work determines whether training data is usable or noise.
- Prompt Engineering
- Specification
- LLM Training
Platform
- Secure contributor IDE with strict workflow controls
- Task creation, review, and dataset validation in one place
- Contributor evaluation and engineering quality gates
- Built for scale across languages and task types
Tech Stack
A FastAPI backend built on clean architecture, deployed on AWS with deep automation, AI evaluation, and production-grade observability.
Core & API
- FastAPI
- Python
- Pydantic
- Clean Architecture
Data
- PostgreSQL
- SQLAlchemy
- JSONB
- Alembic
Cloud & Infra
- AWS EC2
- AWS CodeBuild
- S3
- Secrets Manager
- SSM
- Docker
AI
- OpenRouter
- LLM Evaluation
- Cost Calculation
Auth & Security
- JWT (python-jose)
- RBAC
- bcrypt / passlib
Async & Jobs
- APScheduler
- httpx
- tenacity
- cachetools
Observability
- OpenTelemetry
- Grafana LGTM
- structlog
Integrations
- GitHub API
- Slack SDK
- aiosmtplib