Skip to content
View atillacanuslu's full-sized avatar

Block or report atillacanuslu

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
atillacanuslu/README.md

Atilla Can Uslu

Software engineer — I build AI systems and the infrastructure underneath them.

RAG pipelines, agent tooling, and serverless backends that run in production for real users. USC ECE '25, currently shipping education-technology integrations at MemorySync and building a team-knowledge platform on Azure.

Redmond, WA  ·  atillacanuslu@gmail.com  ·  LinkedIn


MindSync — ask your teammates when they're not around

Every engineer carries context nobody wrote down. MindSync turns a person's own GitHub history and AI coding sessions into a mind their teammates can question — from the web or from inside Claude Code. It answers in their voice with citations, or says "I don't know" and routes the question to the real person.

Two minds asked the same question: one answers with citations, the other declines and offers to escalate to the real person

Two minds, same question. One cites the PRs it drew from; the other has no approved material and declines rather than guessing. Sample data.

What's interesting about it

  • Hybrid retrieval — BM25 + 1536-dim vectors fused with reciprocal rank fusion over Azure AI Search, then re-weighted by source trust × freshness × relevance.
  • It refuses. An empirically measured relevance gate means a thin-evidence question gets a decline and an escalation to the real person, not a fabrication. Every ask is logged with its evidence score so the threshold is tuned from data.
  • Privacy is a test, not a promise. Session capture splits your turns from the assistant's on your machine — guarded by an allowlist test that plants markers in assistant text, tool output, subagent prompts, and compaction summaries and asserts zero survive.
  • Consent by construction. Nothing enters a person's mind unapproved, and nobody approves material into someone else's.

Azure Functions Cosmos DB Azure AI Search Azure OpenAI Bicep TypeScript MCP GitHub Apps

Live app  ·  Code + architecture write-up  ·  ~$9.50/month to run  ·  176 tests


Experience

Software Engineer — MemorySync LLC · Sep 2025 – Present · Redmond, WA
Built EduSync, a cross-platform writing-analytics product, end to end: ingestion and text-processing pipelines that parse student work across Canvas LMS (LTI 1.3), Microsoft Word, and Google Docs; a serverless multi-tenant backend on Firebase Cloud Functions with JWT auth. Published on the Canvas App Center, Microsoft AppSource, and Google Workspace Marketplace — active pilot at Seattle University.

Software Developer Intern — TaleWorlds Entertainment · Summer 2023 · Ankara, Türkiye
Shipped AI and gameplay features in C++/C# for Mount & Blade II: Bannerlord — optimized NPC pathfinding for large-scale battles and built a dynamic random-encounter mission system.

Software Engineering Intern — Credit Registration Office · Summer 2022 · Istanbul, Türkiye
Engineered C++ real-time, high-volume data-processing pipelines for credit-scoring systems, cutting latency and improving reliability under heavy load.


Other work

memorysync-ai — AI-powered personal file assistant. A local-document RAG system with OpenAI embeddings over Qdrant, a multi-format processing pipeline (PDF, DOCX, images with OCR), and an intent-classification layer routing across 8 categories. Python Cloud Functions backend, Flutter macOS client with live file-system monitoring.


Toolbox

Languages — Python · C · C++ · C# · JavaScript · TypeScript · Dart

AI / LLM — RAG pipelines · hybrid search (vector + BM25, RRF) · Azure AI Search · Qdrant · OpenAI + Azure embeddings · semantic chunking · grounded generation with citations · retrieval evaluation and threshold tuning · Model Context Protocol (MCP) · structured outputs · OCR

Cloud & Backend — Azure (Functions, Static Web Apps, Cosmos DB, AI Search, Key Vault, Foundry) · Google Cloud Functions · Firebase · Node.js · serverless · queue-based async pipelines · REST APIs · JWT auth · managed identity · Bicep / IaC · Docker

Integrations — LTI 1.3 · Canvas LMS API · Microsoft Graph · Google Workspace APIs · GitHub Apps and webhooks · Flutter

AI-assisted development — Claude Code · Cursor · Copilot, on a verify-before-trust workflow: everything generated gets reviewed and tested before it ships.


In progress

  • Microsoft Certified: Azure AI Apps and Agents Developer Associate (AI-103)
  • AWS Certified AI Practitioner

Open to software engineering roles — AI/ML infrastructure, backend, or full-stack. Reach me at atillacanuslu@gmail.com.

Popular repositories Loading

  1. speechify speechify Public

  2. mindsync-plugin mindsync-plugin Public

    Ask your teammates when they're not around — each person gets a digital mind built from their own approved words, answering with citations or declining honestly. Claude Code plugin + system overview.

    JavaScript

  3. atillacanuslu atillacanuslu Public

    About me — profile README

  4. chat_rag chat_rag Public

    Forked from MurselTasgin/chat_rag

    Python