Mert Sırakaya
Software Developer
Istanbul, Türkiye contact@tomris.dev
https://www.tomris.dev
Profiles
fswair
swair
mert-sirakaya
edulayer
Intro
Applied AI Engineer & Python Developer focusing on architecting production-grade agentic systems with modern, typed
Python. I build reliable agents using Pydantic AI and also contribute upstream with merged PRs and multiple commits
to the core project. My work emphasizes AI observability (Logfire), rigorous AI evaluation (Pydantic Evals), structured
outputs, and scalable backend integration. I enjoy sharing my technical insights through video tutorials; you can view
my deep-dive on C++ Iterators as an example of my teaching style. I actively contribute to the open-source ecosystem
by publishing reusable packages to PyPI.
Current Focus
I bridge the gap between LLMs and production by integrating automated AI evaluations into TDD workflows as a first-
class testing mechanism. In my VenusAI framework, I engineered runtime recovery and self-healing capabilities,
enabling tools to self-correct via context-aware retries and dynamic dependency injection. I'm specializing in
evaluation-driven agent systems and runtime reliability. Parallel to this, I apply advanced prompt optimization
techniques using DSPy, GEPA, and MiProv2. My focus remains on advancing agent runtime infrastructure for fully
production-ready systems.
Experience
Contract
Python Engineer
Jun 2022-Present
Remote
Delivered end-to-end Python solutions focusing on automation and Web3 integrations. Developed asynchronous
Telegram bots for real-time notifications, built robust ETL pipelines for continuous data aggregation, and integrated
secure crypto payment gateways for DEX interfaces. Also, I have used my own ORM library (MentoDB) for database
interactions in my client projects.
Projects
VOWEL
3k+ Downloads
Nov 2025-Present
https://pypi.org/project/vowel
I built declarative YAML specification language and framework for testing and evaluating Python functions with AI-
powered eval spec auto-generation. It supports pytest patterns such as fixtures, parametrization. Generating and
running AI-generated evaluations took under eight seconds and costs about half a cent per function for eval generation.
evals, testing, metrics, pydantic-evals, yaml spec
VenusAI
4.5k Downloads
May 2025-Present
https://pypi.org/project/venai
I'm developing an extended Pydantic AI fork focused on building more capable AI agents — adding dynamic tool
management, error-safe and self-healing tool execution, sandboxed execution, multi-backend caching, and seamless
MCP integration.
agents, pydantic-ai, dependency injection, self-healing & autofix, mcp
MentoDB
5k+ Downloads
Jun 2022
https://pypi.org/project/mentodb
I built an ORM mount for SQL engines that supports various output formats, including Pandas DataFrames and JSON-
based structures, provides Pydantic integration, and allows direct access to raw SQL connections. I have used MentoDB
in freelance client projects for local database integrations.
orm, sql, pandas, pydantic
Education
Istanbul University
Web Design and Coding
GPA: 3.41
Aug 2023-Jan 2026
Associate (OEF)
Skills
Python
oop, fastapi, scraping &
automation, backend,
pypi, typed, pep8, async
programming
PydanticAI
agents, graphs, tool
calling, embeddings, ai
evals
DSpy
signatures, prediction,
llms, gepa
Logfire
observability,
monitoring, debugging,
ai observability
Large Language
Models
groq, ollama, lm studio,
google ai studio, ocr,
rag, embedding, prompt
optimization
Vector Databases
chromadb, rag,
embeddings, chunking
n8n (AI Workflows)
n8n, ai workflows,
workflow automations
Pydantic
data validation, json
schema, serialization
Test Driven
Development
tdd, pytest, mocks,
fixtures, unit testing
CI/CD
aws, cloudwatch,
lambda, nginx, docker,
github actions, vercel
C++
oop, stl, generics,
c++20, modern cpp,
iterators
SQL
orm, sqlite, sqlalchemy
MongoDB
nosql, aggregating, data
modeling
Certifications
The C++ Programming Language
Plepa Training & Consulting
Jul 2023-Jan 2024
Gained a solid foundation in Modern C++ (OOP, STL,
Templates) to handle low-level logic and multi-language
development tasks.
MongoDB Python Developer Path
MongoDB Academia
Mar 2025
validate
Completed the official developer path focusing on data
modeling, aggregation pipelines, and indexing strategies.
Languages
Turkish
Native Speaker
English
Business English
Feel free to contact with me from contact@tomris.dev