Skip to content
ARAnas Rasheed

Open to interesting problems

AI Engineer · MOD Ventures · Karachi

Anas Rasheed

Shipping RAG pipelines, agents, and LLM products that survive real users.

I build production AI systems on a solid backend core. 5+ years with Node.js and TypeScript, now focused on RAG pipelines, multi-agent architectures, and LLM integrations that run in real products, not demos.

About

I'm an AI engineer from Karachi with a backend foundation built over years of shipping APIs, real-time systems, and payment integrations. That foundation is what lets me take GenAI work seriously: RAG pipelines, multi-agent systems, and LLM integrations that go from architecture to production, not just prototypes.

I've led backend on multi-tenant AI platforms, built natural language to SQL agents for enterprise databases, researched NPC intelligence for game AI, and shipped self-improving automation systems. I care about AI systems that hold up under real usage: clean architecture, honest error handling, and features that keep working when clients depend on them.

Currently

Software Engineer · MOD Ventures

Karachi, Pakistan

Projects

Colaraz

2026 — Present · Backend Lead

The AI/agent backend for a Pakistan student-admissions platform. Matches students to university programs and scholarships, answers questions through an AI support agent and a RAG FAQ chatbot, and runs scraping and evaluation pipelines — orchestrating Claude, Gemini, and OpenAI behind a fallback chain.

FeaturedNestJSLangChainPostgreSQLRAGMulti-LLMBullMQ

Cold Email Automation Platform

2026 · Solo Builder

A self-learning email platform with multi-powered agents: an optimizer that analyzes campaign data and extracts winning patterns, and a writer that generates the next batch from those updated rules.

FeaturedClaude APIAgentsNode.jsAutomation

ModTalk

2023 — 2024 · Backend Lead

A multi-tenant RAG chatbot platform where each business gets an isolated AI assistant trained on its own data. Handled full backend architecture, LangChain integration, model routing, and vector database setup.

FeaturedNestJSLangChainRAGVector DBMulti-tenant

SenseQL

2025 · Backend Lead

A natural language to SQL multi-agent system supporting BigQuery, PostgreSQL, and MySQL. Solved LLM context limits on large databases with a schema filtering layer that passes only relevant tables and columns to the model.

FeaturedTypeScriptMulti-agentBigQueryPostgreSQLLLM

RawPowerGames NPC AI

2025 — 2026 · AI Researcher

AI research for a UK game studio building an AI faction leader that 10 to 50 players could privately converse with. Modeled NPC personalities with OCEAN traits, designed selective vector memory in Pinecone for key game events, built a Neo4j knowledge graph for faction relationships and game state, and researched reinforcement learning for dynamic NPC behavior.

LangChainNeo4jPineconeGeminiGame AIPython

Now

What I'm focused on right now.

Work

  • Colaraz

    Building the AI/agent backend for a Pakistan student-admissions platform — program and scholarship recommendation, an AI support agent, and a RAG FAQ chatbot.

  • Recommendation quality

    Running an LLM-judge evaluation loop that scores recommendations and proposes prompt rewrites to A/B test.

Learning

  • Convex

    Exploring reactive backend platforms alongside Next.js for real-time product architectures.

Life

  • Training

    In the gym 5 to 6 days a week working toward a weight gain goal.

Experience

  1. Software Engineer · MOD Ventures

    2023Present

    Karachi, Pakistan

    Led backend on AI products across multiple client platforms, from multi-tenant RAG chatbots to agent-based automation systems.

    • Led backend on ModTalk, a multi-tenant RAG chatbot platform serving 15+ businesses with isolated, client-specific AI assistants.
    • Led backend on SenseQL, a natural language to SQL multi-agent system for BigQuery, PostgreSQL, and MySQL, cutting query turnaround from 10-20 minutes to under a minute for enterprise clients.
    • Contributed AI research for a UK game studio: OCEAN-based NPC personality models, Neo4j knowledge graph for faction memory, selective Pinecone vector memory, and reinforcement learning research for dynamic NPC behavior.
    • Built a production support agent system for an EdTech platform with intent classification, a multi-handler architecture, and LangSmith-based evaluation.
    • Shipped an agent-based content generation system with real-time UI indicators, plus media migration to DigitalOcean Spaces.
  2. Junior Software Engineer · Protech Source

    20212022

    Karachi, Pakistan

    Full-stack work on client platforms including a veterinary appointment system and a chef booking product.

    • Built Equigate, a veterinary platform with real-time chat, Stripe Connect payments, and automated appointment reminders on AWS.
    • Developed iFedYou, a chef booking platform with recurring Stripe subscriptions and a background service for tracking chef working hours.

Skills

Languages
TypeScriptJavaScriptPythonJavaSQL
AI / GenAI
LangChainLangSmithRAG PipelinesMulti-agent SystemsVector DatabasesNeo4jPineconePrompt Engineering
Backend
Node.jsNestJSExpressPostgreSQLMySQLMongoDBRedis
Cloud & Tooling
AWSGCPDigitalOceanDockerCI/CDStripen8n

Get in touch

Always happy to talk about AI engineering, agent systems, or the backend architecture that holds them together. Email is the fastest way to reach me.