Filip Mijic
AI Automation & Integration Engineer — n8n, AI Agents, RAG
New to Contra
Filip is ready for their next project!
Multi-Agent Orchestrator Development
0
0
An open-source Python RAG app. Documents (PDF/DOCX/MD/TXT) are loaded, split into paragraph-aware overlapping chunks, embedded, and stored in a ChromaDB vector store. At query time the question is embedded, the closest passages are retrieved via cosine top-K search, injected into a grounding-and-refusal system prompt, and the answer is streamed back with its source chunks shown. LLM and embedding providers are pluggable (Groq/OpenAI and local sentence-transformers/OpenAI) via .env, and it ships with both a Streamlit UI and a CLI. An evaluation harness measures retrieval hit-rate and answer accuracy on a labelled Q/A set.
0
9
AI resume builder with RAG chat — shipped to production, $0 recurring infra.
I design and build lean, production-grade AI products. My latest, kairesume (live at kairesume.fit (http://kairesume.fit)), tailors resumes to any job description, scores ATS keyword match, drafts cover letters, and answers user questions through a retrieval-augmented chat assistant grounded in real product docs — so it stays accurate.
What I bring to your project:
RAG systems — pgvector, embeddings, chunking, retrieval that doesn't hallucinate
Full-stack AI apps — Next.js 14 + TypeScript, Supabase, Groq/LLM integration, streaming chat
Cost-efficient architecture — I shipped this whole system on a $0 infra budget by making sharp tradeoffs
Production workflow automation — n8n ingestion pipelines
If you need an AI feature built right — RAG chat, document tooling, or an LLM-powered MVP — let's talk. Available for freelance and contract work now.
1
38
Automated RAG vector store sync (n8n + Supabase pgvector)
The problem: a retrieval-augmented chatbot whose knowledge base drifts out of date the moment the source docs change — unless someone remembers to manually re-embed everything.
What I built: a fully automated reindex pipeline in n8n that keeps the vector store current with zero manual steps.
How it works:
Triggers on a daily schedule and on a deploy webhook
Fetches source markdown, chunks it on heading boundaries with overlap to preserve context
Embeds in batches via a Supabase Edge Function (384-dim)
Bulk UPSERTs vectors into Postgres with pgvector
Removes stale chunks from deleted or shortened docs automatically
Why it's solid:
Idempotent — safe to re-run any number of times, no duplicates, no drift
Runs entirely on free tiers ($0 infra)
Retry-on-fail on network calls + a dedicated error workflow for alerts
I design and ship production automations like this — RAG pipelines, data sync, internal tooling, and API integrations. If you're building anything with AI retrieval or you've got a repetitive data process worth automating, send me a message and let's scope it.
0
22