<MI/>
available_for_work

Mohammed IkramAshrafi.

Python · GenAI · RAG Engineer

I build production AI systems that ship — not demos that crash. Python, FastAPI, Django, and LLM pipelines from idea to deploy.

trusted stackGroqQdrantPineconeLangChainFastAPILlama 3.3
ikram@portfolio ~ %
  • Python
  • FastAPI
  • Django
  • LangChain
  • Llama 3.3
  • Qdrant
  • Pinecone
  • Groq
  • React
  • TypeScript
  • PostgreSQL
  • Docker
02Featured Work

Projects I've shipped.

Production-ready systems showcasing GenAI pipelines, full-stack engineering, and scalable architecture.

Headline · Reflective Memory Intelligence · 2026

EchoVault AI

Reflective memory intelligence platform. Temporal Emotional RAG that chunks by feeling, embeds per-user vector spaces, and answers grounded — every claim cited, no impersonation.

LiveSolo build · 2026Open sourceMIT licensed
FastAPIPython 3.13QdrantGroqLLaMA-3BGE Embeddings
Read case studyLive demo
Marketing hero — preserving memories through ethical AI.
02 / 03
live

Enterprise AI Platform

GenAI RAG Chatbot

Enterprise-grade AI chatbot with Retrieval-Augmented Generation using Llama 3.3-70B via Groq API. FastAPI backend with JWT and Google OAuth, role-based access, and a complete RAG pipeline for PDF, DOCX, TXT, and Markdown.

p95 latency

<200ms

model

Llama 3.3-70B

status

production

PythonFastAPILangChainPinecone
View on GitHub
03 / 03
live

Complaint Management Platform

Anti-Ragging System

Comprehensive complaint management system with Django backend and React frontend. Four distinct user roles (Admin, Principal, Squad, Students) with granular permissions and real-time updates.

roles

4 / RBAC

notifs

real-time

status

production

PythonDjangoDRFReact
View on GitHub
view all case studies
03Currently

What I'm shipping this week.

Updated regularly. No stale content.

  • shippingEchoVault AI — reflective memory intelligence with grounded citations. Live at echovaultai.me.
  • buildingPhase 1 of EchoVault: Clerk auth, per-user Qdrant namespaces, managed Postgres.
  • learningvLLM, model quantisation, and serving LoRA adapters at low latency.
  • open toPython / AI Engineer roles, remote or India-based.
04Let's Connect

Let's build something amazing.

Open to Python / AI Engineer roles, freelance work, and collaborations on intelligent products.

Start a conversation ikramshariff2005@gmail.com