I am  . In order to achieve something great, I believe working with a team with strong chemistry is key. I adapt well to adverse situations while staying positive.
I am looking for opportunities that challenge me to bring out the best in myself and drive impactful change in the world. I also want to explore the uncharted. |
[SYSTEM_IDENTITY]
NAME = Ilhan Mansij Jidda
CALLSIGN = Cheese-Cheese
ROLE = Software Engineer // AI & Distributed Systems
[CORE_DIRECTIVES]
> Full-Stack & Systems Architecture: Architecting production backends with FastAPI, Flask, and Next.js.
> Real-Time Voice AI: Engineering multi-agent spoken pipelines using LiveKit Agents, Deepgram Nova-3, and Murf Falcon TTS.
> Computer Vision & Multimodal: Building U-Net tissue segmentation, MobileNetV2 classifiers, and Gemini/LLaMa/GPT-OSS multimodal reasoning engines.
> Hackathon-Tested: Engineered clinical diagnostics (MIT Hackathon) and bilingual voice agents (Voice for Bharat).
> Engineering Standard: Finite State Machine (FSM) control flow, strict output validation, and zero hallucinations.
LANGUAGES |
|
AI / LLM & CV |
|
BACKEND & WEB |
|
DATABASES |
|
DEVOPS & TOOLS |
|
•
AI-Powered Resume Review & Real-Time Voice Interview Platform
Smart Platform for Adversarial Readiness & Technical Assessment. Compares resumes against job descriptions using Groq-hosted LLMs, PyMuPDF extraction, and NVIDIA Nemotron OCR fallback for scanned pages. Key Architecture: Real-time voice interview engine (Deepgram STT) with Finite State Machine (FSM) orchestration for predictable dataflow, repository coverage reporting, and offline test suites. 
•
[SOURCE_REPO]
•
[LIVE_DEMO]
|
•
Animated Retro CRT TV GitHub Statistics Widget & GitHub Action
Zero-runtime-dependency GitHub Action dynamically generating customizable, animated CRT television SVGs with cycling channels for live system stats, language breakdowns, repository spotlight, and activity radar. Features a live interactive studio for real-time theme customization. Key Architecture: TypeScript/Node.js rendering engine bundling pure SVG output with embedded VT323 pixel typography, CSS @keyframes phosphor scanlines/noise textures, and GitHub GraphQL API integration (GitHub Camo compatible). 
•
[SOURCE_REPO]
•
[LIVE_PREVIEW]
•
[MARKETPLACE]
|
•
Real-Time Spoken English Voice Tutor (Voice for Bharat Challenge 2026)
Engineered a multi-agent spoken English tutor across Murf Falcon streaming TTS, Deepgram Nova-3 STT, and Gemini 2.5 on LiveKit Agents, backed by Next.js and outbound SIP telephony. Key Architecture: Integrated live dictionary & grammar REST APIs as LLM tools with spoken fail-soft fallbacks, human teacher escalation dashboards, and a 22-test pytest suite. 
•
[SOURCE_REPO]
•
[TECH_ARTICLE]
|
•
Clinical Support System for Chronic Wound Diagnostics (MIT Hackathon)
Dual-model computer vision pipeline combining U-Net semantic segmentation and MobileNetV2 tissue classification with reference-patch calibration converting pixel measurements to real wound surface area (cm²). Key Architecture: Flask REST API backend with Streamlit multi-role clinical dashboard (Doctor/Patient portals) and automated PDF diagnostic reporting. 
•
[SOURCE_REPO]
|
> TELEMETRY_BROADCAST: Like this CRT telemetry monitor? Customize & deploy your unit •
[LIVE_STUDIO] • [DEPLOY_CATHODERAYSTATS]
[EOF] TRANSMISSION TERMINATED // STATUS 200 OK // HOST: CHEESE-CHEESE