Ifrah
Ashraf
Full Stack Developer
Engineer driven by curiosity and first principles. I build full-stack systems, autonomous AI agents, and anything that solves real problems from the ground up. Started with C, never stopped exploring.
What I bring
Skills & Expertise
I don't just write code — I architect solutions and think from first principles.
What I build
Engineering
- Full Stack Development
- AI / Agent Systems
- Backend Architecture
- API Design & Integration
- Database Design
- Cloud Infrastructure
How I think
Approach
- First Principles Thinking
- Systems Design
- Technical Writing
- Open Source Contribution
- Performance Engineering
What I use
Tech Stack
Where I've worked
Work Experience
Real-world engineering at fast-moving companies — shipped code, solved hard problems.
Agentic AI Data Engineer Intern
Harman Connected Services
Working in the Data Engineering team building a full-fledged multi-agent system using LangGraph. Focus on the supervisor layer — designing how the supervisor coordinates multiple sub-agents, manages handoffs, and incorporates human-in-the-loop for critical decisions.
Key Contributions
- Designing supervisor coordination logic for multi-agent pipelines using LangGraph
- Integrating MCP (Model Context Protocol) server for agent tool access
- Implementing human-in-the-loop checkpoints for critical data decisions
- Building agentic workflows that automate end-to-end data engineering tasks
Full Stack Developer Intern
Poster.fun
Built and shipped production features for a design-generation platform. Owned the AI image pipeline and authentication infrastructure end-to-end.
Key Contributions
- Reduced AI image API response time from 2 min → 15 sec by offloading storage to async Redis jobs via BullMQ
- Integrated social logins via Privy — X (Twitter) and OAuth services
- Architected async job queue system for non-blocking AI generation workflows
- Shipped full-stack features across frontend and backend in a fast-paced startup
What I've built
Featured Projects
Each project is a story of solving a real problem — with outcomes, learnings, and impact.
Multithreaded Web Scraper
Concurrent data extraction at scale
Built from scratch using Go's concurrency model to scrape and process thousands of URLs simultaneously with zero race conditions.
Key Outcomes
- 10x faster than single-threaded alternatives
- Zero-downtime concurrent processing
- Handles 1000+ URLs in parallel
What I Learned
Mastered goroutines, channels, and WaitGroups. Learned to design race-condition-free concurrent systems.
Smart India Hackathon — School Resource Allocator
Algorithm-driven resource standardization
Designed and built the core algorithm that dynamically allocates educational resources to standardize schools across different tier categories.
Key Outcomes
- Submitted to national-level competition
- Dynamic allocation with constraint satisfaction
- Reduced manual effort by ~80% in resource planning
What I Learned
Deep dive into constraint-based optimization, greedy algorithms, and building systems with real government-scale impact.
RAG System from Scratch
Context-aware retrieval pipeline
Built a RAG pipeline by extracting PDF data page-wise, cleaning it, and adding metadata like headings and page numbers. Structured the flow using a Medallion approach before generating embeddings and storing them for retrieval.
Key Outcomes
- Structured Bronze → Silver → Gold style data flow
- Generated embeddings using voyage-3.5
- Stored and queried vectors using Qdrant DB
What I Learned
Understood how data cleaning and chunking impact retrieval quality and how to structure a RAG pipeline end-to-end.
Milestones
Achievements & Work
Things I've built, shipped, competed in, and contributed to.
Smart India Hackathon Finalist
National-Level Competition · 2024
Built and presented a dynamic resource allocation algorithm for standardizing government schools. Competed at national level.
Open Source Contributor
GitHub Projects · 2025 – Present
Actively contributing to production open-source projects used by thousands of developers worldwide.
Contributions
AI Agent Builder
Autonomous Systems · 2024 – Present
Developing autonomous AI agents and LLM-powered tools that go beyond basic chatbots reasoning, planning, and taking actions.
Full Stack Engineer
Production Systems · 2022 – Present
Shipped multiple production ready full-stack applications spanning real time systems, APIs, and modern frontends.
Thoughts & Writing
The Blog
Engineering deep dives, lessons from building, and ideas worth sharing.
From Documents to Answers: How RAG Works
Learned how basic cosine similarity used in vector search