# Bogdan-Alexandru Truța

> AI Engineer at ERNI and co-founder of AlphaProve, based in Cluj-Napoca,
> Romania. Builds production AI systems, quantitative trading infrastructure,
> and autonomous-agent tooling. Cares most about correctness, isolation, and
> systems that keep running when nobody is watching.

Contact: bogdan@bogdantruta.com · https://bogdantruta.com

## Current roles

- AI Engineer, ERNI (May 2026 to present, Cluj-Napoca): grounded LLM
  generation with source-traceable outputs, hybrid lexical + vector retrieval,
  and LLM-as-judge harnesses that score generation quality.
- Co-Founder, AlphaProve (Jan 2026 to present): a deployed multi-tenant
  quantitative backtesting SaaS, in private beta. Designed the AI loop that
  turns a plain-English strategy description into executable code validated
  by backtest, and built the backtesting engine: tick-precision exchange
  simulation with L2 order-book fills and ML-driven walk-forward validation.
- Software Engineering Intern, NTT DATA (Dec 2025 to Mar 2026): led a
  6-developer Agile team that delivered a hotel-booking REST API in Java and
  Spring Boot with JWT/RBAC security, shipped through GitHub Actions CI/CD to
  a Dockerized Linux VPS.

## Projects

- [AlphaProve](https://alphaprove.com): multi-tenant quantitative backtesting
  SaaS with tick-precision exchange simulation and walk-forward validation.
- [eGata](https://github.com/Bogzx/eGata): AI agent that walks Romanian
  citizens through municipal procedures over voice, text and telephone.
  Retrieval-augmented procedure matching, LaTeX-signed PDFs, and a
  tamper-evident hash-chain audit ledger. Built at Cluj Hackathon 2026.
- [GOSHA Jobs](https://gosha.bogdantruta.com): deployed job-matching platform
  that ranks openings against a candidate's CV using sentence-transformer
  embeddings and Rocchio relevance feedback. Used by 20+ students.
  [Source](https://github.com/Bogzx/gosha-jobs).
- TradingStats: quantitative research framework. Hypotheses are registered
  before evaluation, look-ahead bias and leakage are detected automatically,
  and negative results are recorded alongside positive ones.
- [CS2 Weekly Item Journal](https://github.com/Bogzx/CS-Weekly-Item-Journal):
  computer-vision pipeline that reads Counter-Strike 2 weekly item drops from
  screenshots with a custom-trained YOLO model and OCR.

## Technical focus

- Languages: Python, TypeScript, JavaScript, Java, C, C++, Rust, SQL
- AI and ML: LLM integration and agent systems, MCP, Langfuse, PyTorch,
  YOLO, OpenCV, sentence-transformers
- Backend: FastAPI, Celery, Spring Boot, Node.js, Hono, Next.js, React
- Infrastructure and data: Docker, GitHub Actions CI/CD, PostgreSQL,
  TimescaleDB, pgvector, Redis, SQLite
- Systems and security: seccomp-BPF sandboxing, Linux namespaces, JWT,
  Spring Security

## Education and recognition

- BSc Computer Science, Babeș-Bolyai University, Cluj-Napoca (expected 2028)
- 1st place, App Development, PoliHack 2025
- 2nd place, AI App Development, PoliHack 2026
- 4th place, AI/Data Science, Cloudflight Coding Contest 2025
- National Robotics Champion, NEXTLAB.TECH 2021

## Writing

Essays at https://bogdantruta.com/blog/ (RSS: https://bogdantruta.com/feed.xml).


## Pages

- [Home](https://bogdantruta.com/): projects as an interactive 3D commit graph
- [Experience](https://bogdantruta.com/experience.html): work history
- [CV (PDF)](https://bogdantruta.com/BogdanTruta-CV.pdf)